1

Music Restoration Handbook

Early Disc Records to Today’s LPs

Marc S. Hildebrant

5.0 Edition
 2

Copyright © 2022 by Marc S. Hildebrant

All rights reserved. This book or any portion thereof
may not be reproduced or used in any manner whatsoever
without the express written permission of the publisher
except for the use of brief quotations in a book review.

Printed in the United States of America

Sixth Printing, 2022

5.0 Edition
HCLLC
P.O. Box 68
West Dennis, Massachusetts 02670
3

Dedication

To My Wife who continues to inspire me to achieve so much in my Life.

Marc Hildebrant
4

The Author and One of his Grandchildren Enjoying Restored Music!

 5

Table of Contents
1.0 Introduction ......................................................................................................... 12
1.1 About the Author and Handbook ..................................................................... 12
1.2 The Layout of this Handbook .......................................................................... 14
1.3 Record Types Addressed in this Handbook .................................................... 15
Basic Music Restoration Techniques........................................................................... 16
2.0 Definitions of Restoration & Enhancement ........................................................ 17
2.1 Restoration ....................................................................................................... 17
2.2 Enhancement .................................................................................................... 17
3.0 Motivation for Restoration of Music .................................................................. 19
3.1 Records Wear Out ............................................................................................ 19
3.2 Why Restoration of Music is Important and Worthwhile ............................... 19
3.2.1 Advantage of Using the Correct Playback Equipment for Records ......... 21
3.3 Early Music Releases from CD Restorations & Digital Downloads ............... 21
4.0 Tools Needed for Audio Restoration .................................................................. 22
4.1 Restoration Software ........................................................................................ 25
4.2 Computer Operating Systems Software ........................................................... 25
4.3 Audio Restoration Hardware ........................................................................... 25
4.3.1 Computer .................................................................................................. 25
4.3.2 Analog to Digital & Digital to Analog Converters .................................. 26
4.3.3 Listen while Recording............................................................................. 27
4.3.4 Turntable & Cartridge .............................................................................. 28
4.3.5 Preamplifier and Gain Control ................................................................. 28
4.3.6 Playback of Music during the Restoration Process .................................. 30
4.3.7 Setting Windows Privacy for Microphone ............................................... 31
4.3.8 Control of Sample Rate for Converters Internal & External .................... 32
4.4 Overview of DCart10 Software Operation for Restoration ......................... 33
4.4.1 The Operation of the Software ................................................................. 33
4.4.2 Multifilters ................................................................................................ 36
4.4.3 Music Storage and CD Creation ............................................................... 36
5.0 Music File Organization ...................................................................................... 37
5.1 Music File Type Used in The Handbook ......................................................... 38
6

5.2 General Organization Method .......................................................................... 38

5.3 Specific Folder Naming within Major Locations ............................................ 39
5.3.1 Original Source Music Recordings Layout .............................................. 39
5.3.2 Restoration Files ....................................................................................... 40
5.3.3 Restored Music Files ................................................................................ 41
5.4 Summary .......................................................................................................... 41
6.0 Music Equalization & Removal .......................................................................... 42
6.1 Overview .......................................................................................................... 42
6.2 Definition of Terms .......................................................................................... 42
6.3 DCart10 Equalization Tools ............................................................................. 44
6.4 How to Produce a Flat Recording from a Non-Flat Preamplifier .................... 47
6.4.1 Two Types of RIAA EQ Filters are Present, Which One to Use? ........... 47
6.4.2 The Types of EQ that can be Removed .................................................... 48
6.5 How to Determine if the Record was Acoustical or Electrically Recorded .... 48
6.6 The Correct Equalization for a 78 Electric RPM Record ................................ 52
6.7 Note Regarding Pre-Emphasis on 78 Electric Records ................................... 54
6.8 Special EQ for Edison Electric Recordings ..................................................... 54
6.9 EQ for LP and 45 Records .............................................................................. 55
7.0 Transcribing Records .......................................................................................... 56
7.1 Definition of Terms .......................................................................................... 56
7.2 Equipment Set-up ............................................................................................. 56
7.3 Preamplifier and Turntable Speed .................................................................... 57
7.4 Cartridge and Stylus ......................................................................................... 58
7.5 Channel Selection ............................................................................................. 58
7.6 Sample Rate and Bit Depth Selection .............................................................. 59
7.6.1 Setting the Converters Sample Rate and Bit Depth for Recording ......... 60
7.6.2 Checking the Recording Window............................................................ 64
7.7 Playback System .............................................................................................. 66
7.8 Details of the Transcription Operation............................................................. 67
7.8.1 Clean the Record Surface ......................................................................... 67
7.8.2 Set the Speed to the Value for the Record ................................................ 67
7.8.3 Prepare the Software ................................................................................. 67
7.8.4 Recording of Raw File .............................................................................. 68
7

7.8.5 Creating a Single Raw File ....................................................................... 71

7.8.6 Creating Raw Files from a Multiple Song Recording .............................. 71
7.9 Steps to Create the Original File from the Raw File ....................................... 72
7.9.1 Checking and Correcting the Channel Balance ........................................ 73
7.9.2 Trimming File Length and Checking for Possible Overload ................... 76
7.9.3 Extra Steps to Remove EQ and/or Correct the Playback Speed .............. 79
7.10 Final Preservation as Original File ............................................................... 83
8.0 General Noise Removal Methods & Level Creation .......................................... 84
8.1 Definition of Noise........................................................................................... 84
8.2 View used in DCart10 Software Display ......................................................... 85
8.3 Details of Software Noise Removal Algorithms and Sample Rate ................. 85
8.4 How Much Noise can be removed? ................................................................. 86
8.5 General Noise Pictures ..................................................................................... 86
8.5.1 Magnified Picture of Noise ...................................................................... 87
8.5.2 Use for the Section before the Music Starts ............................................. 88
8.6 Noise Masking Technique Beginning and End ............................................... 90
8.7 General Noise Steps for all Types of Recordings ............................................ 92
8.7.1 Creation of a Level One File .................................................................... 95
8.7.2 Creation of a Level Two File.................................................................. 105
8.9 Specific Multifilters for Records ................................................................... 110
8.10 Creating a Final Restoration File ................................................................ 111
8.11 Method of Keeping Track of Your Restoration Work ............................... 111
9.0 78 Acoustic Noise Removal Methods ............................................................... 114
9.1 Level One Creation ........................................................................................ 114
9.2 Level Two Creation ....................................................................................... 114
9.3 Details of the 78 Acoustic Multifilter Part One ............................................. 115
9.4 Details of the 78 Acoustic Multifilter Part Two ............................................ 119
10.0 78 Electric Noise Removal Methods ............................................................. 123
10.1 Important Note about 78 RPM EQ Values ................................................. 123
10.2 Level One Creation ..................................................................................... 123
10.3 Level Two Creation .................................................................................... 124
10.4 Details of the 78 Electric Multifilter Part One............................................ 124
8

10.5 Details of the 78 Electric Multifilter Part Two ........................................... 129

11.0 Diamond Disc Noise Removal Methods ........................................................ 133
11.1 Level One Creation ..................................................................................... 133
11.2 Level Two Creation .................................................................................... 134
11.3 Details of the Diamond Disc Multifilter Part One. ..................................... 134
11.4 Details of the Diamond Disc Multifilter Part Two ..................................... 139
12.0 LP/45 Noise Removal Methods ..................................................................... 143
12.1 LP/45 Monaural or Stereo Records? .......................................................... 143
12.2 LP Stylus Size ............................................................................................. 144
12.3 Modification Possible to RIAA Curve for Normal and Worn Condition... 144
12.4 Sample Rate for LP & 45RPM Records .................................................... 145
12.5 Level One Creation ..................................................................................... 145
12.6 Level Two Creation .................................................................................... 145
12.7 Details of the LP 45 Multifilter Part One Normal ...................................... 146
12.8 Details of the LP 45 Multifilter Part One Worn ......................................... 148
12.9 Details of the LP 45 Multifilter Part Two ................................................... 152
Advanced Enhancement Methods & Tune Library ................................................... 154
13.0 General Concepts for the Advanced Methods ............................................... 155
13.1 Starting Condition of Files .......................................................................... 155
13.2 Reasons for Enhancement of the Music...................................................... 155
13.3 Details for the Types of Enhancements ...................................................... 156
13.3.1 First Type of Enhancement ................................................................... 156
13.3.2 Second Type of Enhancement .............................................................. 157
13.3.3 Third Type of Enhancement ................................................................. 158
13.3.4 Fourth Type of Enhancement ............................................................... 158
13.3.5 Fifth Type of Enhancement .................................................................. 159
13.4 Examples of Enhancement Multifilters ...................................................... 159
13.5 Saving the Results of Your Enhancement .................................................. 159
14.0 Enhancement First Method Example (Rumble & Hiss) ................................ 160
14.1 Rumble Removal Acoustic Diamond Discs .............................................. 162
14.2 Surface Noise Removal and Final Cleanup Acoustic Diamond Discs ...... 164
15.0 Enhancement Second Method for Various Record Types ............................. 168
15.1 78 RPM & LP\45 Records .......................................................................... 168
9

15.2 Diamond Disc Electric Enhancement ......................................................... 169

15.2.1 Diamond Disc Electric Enhancement Method ..................................... 169
15.2.2 Check for Strong Discrete Rumble Frequencies .................................. 170
15.2.3 Electric Diamond Disc Enhancement Multifilter ................................. 172
16.0 Enhancement Third Method .......................................................................... 176
17.0 Enhancement Fourth Method ......................................................................... 179
17.1 Operation and Flow of the Enhancement Features ..................................... 184
18.0 Enhancement Method Five ............................................................................ 185
18.1 Introduction to this Method ....................................................................... 185
18.2 Description of Correction Method .............................................................. 186
18.2.1 First; Remove Rumble and Generate Source ...................................... 187
18.2.2 Second; Generate Low Frequency Correction File ............................. 189
18.2.3 Third; Generate High Frequency Correction File ............................... 192
18.2.4 Fourth; Merging of High & Low Frequency Files .............................. 193
18.2.5 Fifth; Cleanup / Final Modification..................................................... 195
18.3 Implementation of the Acoustical Correction Process ............................... 198
18.3.1 Create A Working Copy ...................................................................... 198
18.3.2 Create a Source File............................................................................. 198
18.3.3 Create A Low File ............................................................................... 199
18.3.4 Create A High File............................................................................... 201
18.3.5 Create High & Low File ...................................................................... 202
18.3.6 Create the Final Cleanup Version ....................................................... 204
18.4 Conclusion ................................................................................................. 207
19.0 Tune Library and Creating Your Music Playlists .......................................... 208
19.1 DC Tune Library Basics ............................................................................. 208
19.2 Music Folder Layout on your System ........................................................ 210
19.3 Library Headings ........................................................................................ 210
19.4 Importing the Music File Information into the Library .............................. 212
19.4.1 Individual Record Setting ..................................................................... 212
19.4.2 LP Records ........................................................................................... 213
Reference Section ...................................................................................................... 216
20.0 History of Recorded Sound ............................................................................ 217
10

20.1 Beginning of Recorded Sound .................................................................... 217

20.2 Western Electric Laboratory Introduces Electric Recording ...................... 219
20.3 History of the Edison Record Company ..................................................... 219
20.4 Victor Company and Others Business Models ........................................... 221
20.5 New Field of Audio Engineering Created .................................................. 222
20.6 Can Edison Records be played on Victor Phonographs? ........................... 223
21.0 Recording Details for Music .......................................................................... 224
21.1 Physics of Sound ......................................................................................... 224
21.2 Constant-Velocity and Constant-amplitude ................................................ 225
21.3 Constant-Acceleration (Pre-Emphasis) ....................................................... 226
21.4 Early Cutting Grooves on Discs ................................................................. 227
21.5 Playback of Acoustic Era Records .............................................................. 228
21.6 Amplitude Modification with Frequency for Electric Recording............... 228
21.7 Notes on Filter Response ............................................................................ 229
21.8 Early Electric Recording & Playback Conditions ...................................... 230
21.9 Later Recordings ......................................................................................... 231
21.10 Edison Electrical Recording ....................................................................... 232
22.0 Selection of Cartridge and Stylus for Transcription ...................................... 233
22.1 Introduction ................................................................................................. 233
22.2 Cartridge Selection Acoustic and Electric Recording ................................ 233
22.3 Errors in Playback from Stylus Shape for Lateral and Vertical ................. 234
22.3.1 Pierce and Hunt Analysis...................................................................... 234
22.4 Edison Diamond Disc Vertical Modulated Records Stylus Selection ........ 240
22.4.1 Worst Case Calculations for Two Stylus .............................................. 244
22.5 Stylus Selection for 78 RPM Records ........................................................ 246
22.5.1 Introduction .......................................................................................... 246
22.5.2 Worst Case Calculations for Two Stylus ............................................. 247
22.6 Conclusion................................................................................................... 248
23.0 Audio Standards ............................................................................................. 250
23.1 Logarithms, a needed Concept .................................................................... 250
23.2 Early Sound Intensity Measurements, the VU Meter ................................. 251
23.3 Standards Important for Audio Restoration Work ...................................... 253
11

23.3.1 Head Room and Alignment Levels ...................................................... 254

23.4 Diamond Cut Software Level Indicators, VU and Vertical Axis ............... 254
24.0 Correct Recording Level & Number of Bits to Use ...................................... 257
24.1 Details of The Analog to Digital Conversion ............................................. 257
24.2 Signal to Noise Ratio (SNR) for Audio ...................................................... 258
24.3 A to D Noise ............................................................................................... 259
24.4 Measuring Signal to Noise Ratio and Effective Number of Bits ............... 260
24.5 System Example of SINAD ........................................................................ 261
24.6 Conclusion .................................................................................................. 264
25.0 Providing the Needed Preamplifier Gain for Transcription .......................... 266
25.1 Introduction ................................................................................................. 266
25.2 Preamplifier Gain Values............................................................................ 266
26.0 Sample Rate Needed for Restoration ............................................................. 268
26.1 History of Sampling Analog Data .............................................................. 268
26.2 Sampling Theory ......................................................................................... 269
26.3 DSP Needs for Sample Rate ....................................................................... 270
26.4 Caution with Sample Rate Selection in a Converter .................................. 271
27.0 Audio Measurements ..................................................................................... 272
27.1 Measurement of Actual Converter Calibration........................................... 272
27.2 Verification of Sample Rate in Converters ................................................. 273
28.0 Conclusion and Reader Request .................................................................... 275
 12

1.0 Introduction

1.1 About the Author and Handbook

Music is an important part of my life. At an early age I had a phonograph and records
to play and enjoyed hearing the music. I added to my record collection using money earned
from mowing lawns and babysitting. My initial records were classical, and I quickly
developed an interest in all types of music. At times I would buy LP collections of early jazz
music from the 1920’s and enjoyed this unique sound. If I liked the song, it was not important
to me if the music was current or vintage.
At the time that Rock and Roll music was becoming popular my father started buying
very early record players and records. Most of these old records had a cylinder shape with
performers’ names that I had never heard of. I started to help my father sort and file the records
as his collection grew and grew. At this time (1960’s) collecting early Edison and Columbia
machines and records was just beginning to become historically important. Collecting was
still a hobby (not an investment) and the price for the early phonographs and records were
quite low. My father took advantage of this low cost to build up a large collection of this early
technology. At his peak of collecting the number of cylinder records in his collection
approached 2000 along with a couple hundred phonographs. In the afternoons after school let
out, I would listen to and organize the latest batch of his old records. This went on for many
years during which I developed a lasting interest in old music. While attending college I
obtained BSEE and MSEE degrees and I was off to a career in Electrical Engineering.
As happens to many people, my life changed as I married my Wife and we started our
new life together in the early 1970’s. My interest in music continued but our money and time
had to go towards providing for our growing family. Music was still important to me but
collecting old records had to wait.
In the early 1980’s I started to record music using a cassette tape recorder and some
Edison four-minute cylinder records that were located at my parent’s home. A casual
comment was made by my father stating that the sound recorded from a tape player with
automatic level control seemed to sound better than using the original phonograph when
played back. While his comment did not start me towards music restoration it did stay with
me as I wondered if the music could be improved from the record with some modifications
made to the original sound.
During the 1990’s I obtained a modest collection of Edison cylinder records from my
father’s estate and at the same time found a software program from a Company called
 13

Diamond Cut Productions that provided me with abilities to create a digital version of the
song, along with some basic clean-up tools for the music (DCart3 was provided on two floppy
discs at this time).
My career as an electrical engineer had shown me the potential improvements that
DSP (Digital Signal Processing) could provide for improving audio signals. In addition, the
steady increase in the computing power and decrease in cost for desktop computers provided
a means to implement very complex DSP math on music files. My father’s early observation
about improving sound from the use of automatic level control started to provide some music
improvements with the use of the Diamond Cut Productions software. My career continued
as an electrical engineer with some occasional music restoration work.
Around 2010 my career as a full-time electrical engineer changed to retirement
activities and engineering consulting work. I returned to the Diamond Cut Productions
software products and found that the company had made significant improvements to their
early products. I purchased additional computer hardware and created a written journal of my
music restoration work while using their software. I also joined the Diamond Cut Productions
forum and other related audio groups to discuss and learn about music restoration.
During this same time, I met an early record collector with a very large collection of
both 78 RPM records and Edison Diamond Disc records. I took on the task of learning how
to restore records for him from his collection, along with Edison Records and LP’s that I
owned (My collection of Edison records had also been growing).
While I was learning to use the software methods from the Diamond Cut Productions
Company it became clear to me that there were limited resources available to guide you
through the process of restoring music. The help files and owner’s manual that comes with
the software are useful however, much of the needed knowledge is learned by trying a method
and then listening to the results followed by another attempt and another. As I used the
software tools to restore the music, I found that I would have to repeat my work many times
as I was constantly learning new techniques.
Another problem I found in restoring the music was that much of the technical
information about recording and playback of 78 RPM and LP records is found only in out-of-
date books and magazines. Today, the technology of recording records is not needed as the
sound is captured from the recording room and turned into a digital stream of data rather than
a varying analog signal.
I wrote this Handbook so that people with an interest in early music to the latest LPs
could enjoy their non-digital music again with the clear sound as recorded in the music studio
and in turn store that music and play it back using today’s technology. I have included specific
information in this handbook that reflects the many hours of time that I spent learning how to
 14

restore music. I also wanted to introduce you to audio technology so that you would have
confidence in your restoration of music by understanding the reasons for various software
settings.
This handbook will help you to remove a large amount of the wear and tear on your
records and to enjoy their music again as originally recorded. It is expected that you have
either used or will be using Diamond Cut Productions Software and would like to improve
your techniques in the audio restoration process.
I am not an employee of the Diamond Cut Productions Software Company nor was I
paid to write this Handbook by them; rather I found their products to be a useful tool for
restoration and enhancement of music.

1.2 The Layout of this Handbook

This handbook covers a subject that is somewhat difficult to understand. The reason is
that knowing how to restore music does not follow in a linear learning flow. What is meant
by this is that the learning process does not follow learning one subject after another till you
arrive at a final stage of understanding. The restoration process involves several interrelated
subjects to be applied at the same time. As you are learning how to transcribe music you will
also need to understand the history of music, why the various equalization methods were
developed, the correct number of digital bits to represent the music, etc. You really need to
learn quite a variety of interrelated subjects as you start down the road to producing a restored
song. To help in this Learning process the Handbook has been divided into Three Major
Sections.
The first section labeled Basic Music Restoration Techniques covers the beginning
where the music is transcribed from the original source and then converted to a digital music
file. The new digital file will have any distortion in the form of noise removed and a clean
version is available for advanced work or listening.
The next section labeled Advance Enhancement Methods covers additional processing
of this clean version of the music. These advanced topics include frequency modification to
both the values present and the generation of new frequencies. A special section devoted to
the improvement of Acoustically Recorded Edison Diamond Disc Records with techniques I
have pioneered is present in this section.
The third section is a Reference for technical and historic articles about record creation
and playback
 15

1.3 Record Types Addressed in this Handbook

The records that are discussed and restored in this handbook are all types of flat records;
from the early 78 RPM records, Edison Diamond Disc records, later 78 Electric Recorded
Records, 45 RPM and today’s LP records.
While early cylinder records will not be covered, the concepts shown for these early
mechanically recorded records (acoustic recording) could be successfully used to restore
these cylinder recordings with little modification.
Tape recordings are a unique form of storing music that differs from records in
significant ways and the help files in the Diamond Cut Manual covers most of the needed
information to restore this music.
16

Basic Music Restoration Techniques

 17

2.0 Definitions of Restoration & Enhancement

2.1 Restoration

Restoration of music is defined as the effort to return the music from the playback of
the record to the same sound that occurred during the original recording session. If you listen
to the final restored music, it should sound as if you were hearing the desired result the artists
would have intended when they first created the music at the recording studio. The restoration
process should remove the unwanted modifications to the music that have occurred through
the entire process after the recording of the music in the studio, the manufacture of the records,
and then the playback of the music to the speakers. All the distortion and noise from the
limited recording technology used; the scratches on the record, the dirt and damage to the
grooves, and the defects in the playback should be removed and replaced with an exact
version of what was desired at the recording session. This is a rather high bar for the
restoration work but is possible to come very close to achieving this with the high-powered
hardware and software tools available today. For the restoration process the sound is not
modified with new echo or extra bass and treble; rather the goal is to return the music To the
Way it was Recorded.

2.2 Enhancement

Enhancement of music is defined as the effort to add to or subtract from the restored
music frequency content and musical tones in order to achieve the restorer’s desired sound.
Some examples are adding echo or reverberation to the music, creation of a stereo sounding
song from a monaural recording, or other changes that you may want to obtain their desired
sound.
 18

There are many reasons to enhance the restored music and some examples are:

1. A poor-quality recording was originally made from the master recording. All musical
recordings must record the music using the limitations of the technology at the time of
recording. Even today with the current state of the art in recording technology, poor
microphone placement and studio acoustics can still produce a poor master recording.
2. The record producer’s master recording did not capture the music as wanted by the
musicians. The history of recording music details many times that the artist’s desire in
the sound of the final product was not what the producer and record companies’
management wanted. While new mixes of the music cannot be created, some
modifications to the restored version may be closer to the original artist’s desired
result.
3. The record producer modified the sound of the song to agree with the customer’s
playback equipment that was built with the technology of that time. The technology
for the playback of records has improved from the original playback of tin-foil
recordings to today’s high fidelity sound systems. The powerful stereo amplifiers and
massive speakers available today can fill a listening room with a wonderful and
magnificent sound that can satisfy the most critical listener. This potential sound stage
is a far cry from the music playback systems that were used by many 78 RPM record
users in the past. These record producers understood that if their recording sounded
good on an average system at the time the record was made, then sales of the record
could be strong. Therefore, the same record that sounded good in 1935 on a home
system could sound poor today on a hi-fi system. The next example demonstrates that
modification to the music for the customer’s playback equipment may still occur
today.
4. The current practice of using an algorithm to compress the music file (for example
MP3) will introduce losses in the range and frequency content of the restored music.
This common practice to save memory space seems strange and not natural given the
low cost of digital storage, however the fact remains that it occurs and can be
somewhat overcome through various enhancement methods.

These are just some of the reasons that an enhancement to the restored music maybe
be used and will be described in more detail in the section Advanced Enhancement Methods.
 19

3.0 Motivation for Restoration of Music

3.1 Records Wear Out

Time is hard on all of us. A clean LP record when removed from its jacket becomes a
magnetic for dirt. Accidents like bumping the pickup arm while the record is playing usually
creates a scratch on the record during the best part of a song (Murphy’s Law).
The older 78 RPM records were often played with a steel needle or at best a very heavy
sapphire electric pickup. This type of playback for these records would wear down the record
grooves even though these records were made from a tough material (Ground up Rock with
a Shellac Binder).
You will play your favorite records many times and they will wear out. Restoration can
restore much of the original sound from these worn-out records.

3.2 Why Restoration of Music is Important and Worthwhile

The type of music recorded by the record companies has been constantly changing as
the music tastes of the public have in turn changed. For Jazz music much of the early original
music was recorded using acoustical technology and early electrical recording technologies.
The ability to hear this music as it was originally recorded in the studio is just wonderful.
During the years that LPs (Long Playing Vinyl Records) were popular many people
purchased them to hear the new and exciting Rock and Roll music. These LPs were played
and enjoyed many times and developed numerous scratches. Later, as CDs (Compact Disc)
and other Digital Storage techniques became the preferred media, the re-release of original
artist’s songs that were first heard on LPs had high musical expectation for a new noiseless
version of the same songs once heard on these LPs. Instead, the new Digital versions of the
original music seemed to have a harsh sound and did not contain all the songs that were on
your original issued LPs. The common Greatest Hits CD collections seemed to omit your
favorite song! Now you can create a Digital version from your LP’s that will sound as good
or even better then when you first heard the song on the new LP.
The music that was recorded on records from the very beginning of recorded sound up
to the end of the 78 RPM record era represents many unique and original opportunities for
the listener of the restored music. One of the most important is the ability to hear the original
versions of various music styles played by the original artist. If you enjoy jug band music and
 20

have an old recording of Newport News Blues by the Memphis Jug Band, you can hear the
song (after restoration) in its original glory sung with the enthusiasm and phrasing that only
the original recording artists could provide.
For Bing Crosby fan’s you can hear him sing as a young man with the Paul Whiteman
Rhythm Boys’ knowing that he was destined for a long music career. The Dorsey Brothers,
Glenn Miller, and many of the Swing band leaders got their start in early jazz bands in the
early 1920s. When this music is brought back to life, the musical performances sound just
magical.
When early music was recorded in the music studio a single microphone was used to
capture the sound from the performers. A temporary recording made from a very soft material
was then created as the original studio recording from which the actual records would be later
manufactured. Multiple takes could be done but were often limited to a small number as each
new recording required a unique temporary record. This meant that when the musicians came
into the studio, they had to be ready to record and were expected to make a high-quality
balance sound. Multiple takes, remixing, echo added, and voice overs that are common in
today’s music were just not done. The music you heard on the record was what you would
hear when the musicians played during a live concert. In the opinion of many people this type
of music has a natural sound and the artists sounded authentic and realistic. Some would say
that this is quite a contrast to the over produced, artificial sounding music produced today!
Another benefit with a live recording and the use of one recording microphone is that
all the artists playing in the song would hear each other and the total blended sound for the
recording. This gave the artists additional feedback to their own playing which in turn allowed
them to balance the intensity and tone of their music as it was being blended into the
recording.
Many of these types of recordings can only be heard on these early records. While
today’s music has wonderful digital recording with sound delivered with 16 bits and 44.1
thousand samples per second, these original recordings have an important and unique place
for those that enjoy music in all forms.
 21

3.2.1 Advantage of Using the Correct Playback Equipment for Records

Many people play their music from CD’s or other digital sources of music and are not
aware that most analog recorded music has had major signal modifications made to the music
before it was placed into the record grooves. These music modifications were later removed
in the preamplifiers used to playback these disc records. The modifications were standardized
for the LPs and 45s but were not consistent for the 78 RPM or Edison electrical recorded
records. When an electrically recorded 78 RPM record is played on a current audio system
meant for LP records, the music will sound dull and lack the upper frequencies. When the
correct modifications are applied to the older records the muted high frequency parts of the
music can now be heard and the excessive bass is removed. Often the listener of an old record
will be surprised to hear how good they do sound when played back with the correct
equipment.

3.3 Early Music Releases from CD Restorations & Digital Downloads

The major music companies, while having access to many studio masters, have not
produced many re-issues of older material. There are several smaller companies that have
issued restored recordings and while some are very good a large number of the restorations
consist of simply removing the high frequency tones in the recordings to reduce the surface
noise. The removal of these tones reduces the scratchy sound of the record but also makes the
recording sound dull and muted.
Since you have the advantage of restoring the music using your time and effort, the
results that you can produce are as good as you want the restoration to be and can often exceed
commercial music products.
 22

4.0 Tools Needed for Audio Restoration

A workstation for your audio restoration can be as basic or as elaborate as your budget
allows. A guide to the needed equipment will be shown, but you can increase the level of
complexity and features to any desired level. A simple setup can provide very good results,
whereas each incremental improvement to your audio system will cost greater amounts of
money. Pictures of my audio setup are shown in figure 4-1. Basement space was used for my
Music Restoration Studio.

Figure 4-1 Computer, Monitor, and Playback Speakers.

23

Figure 4-2 Record Turntable

Figure 4-3 Preamplifier and Analog Mixer

24

Figure 4-4 Analog to Digital and Digital to Analog Converter

Figure 4-5 Preamp on Speaker with Playback Power Amplifier behind Monitor
 25

4.1 Restoration Software

The software used in this Handbook is from Diamond Cut Productions, Inc. While the
company sells several products the specific one used is DCart version 10.80 Since the
Diamond Cut Productions Company continuously up-dates and improves their products you
can use future software updates from version 10.80 of their software with the examples shown
in this Handbook.
Their software works very well for audio restoration of music due to the large amount
of control over its operation. Because the software offers many adjustments for the use of the
tools this implies that you will need to understand their operation to achieve the best results.
Once their operation has been learned, a large amount of the noise from the recordings can
be removed with the resulting pleasure of clearly hearing the music again.
After the software is installed, you are urged to refer to the owner’s manual and user’s
guide that comes with the software for help with the tools. This Handbook expects that you
have some limited knowledge of the operation of the Diamond Cut Productions Software.
The engineers who developed these software products provided many adjustments and
features in the implementation of their work. The intent of this handbook is to show what has
worked in my journey to restoring music. You are encouraged to experiment and try to
improve on the methods described within. The methods shown do work very well and are the
result of restoring many songs.

4.2 Computer Operating Systems Software

At the time of writing this handbook Windows Professional Version 11 was the
operating system for the computer. The Diamond Cut Productions web site should be
consulted to see if a different operating system software can work for you.

4.3 Audio Restoration Hardware

4.3.1 Computer

The heart of the audio workstation is the computer system. Many desktop systems can
perform the needed software calculations with no problem. Where you can help with
processing speed is to have a large RAM in the computer for system memory. My system
 26

currently has 8 Gigabytes of RAM installed that enables software programs to minimize disk
drive activity which in turn speeds up the program execution time. Disc drives are great for
storing the large music files that are used but are slow for performing the many operations
used in the software.
If possible, the computer should be dedicated to music restoration. In this manner the
need for an Internet connection is reduced along with any potential problems from a software
virus.
A large screen size for the computer monitor helps to display the noise details in the
song and can show all the various software options that are available to use. The screen size
shown in figure 4-1 is 24 inches (diagonal), which works very well for displaying the music
waveform.
Today the cost per byte of disk drive storage is very low. The best workstation storage
management is to store the completed audio files in all their forms (Original, Restoration
Work, and Final Version), duplicated on two separate external disk drives. These disc drives
can be attached to the computer via USB connections. The time spend on your music
restoration work can be long and the original recordings are not always replaceable. Having
two duplicate copies of the audio files is an easy and low-cost method to safely store the
restoration material. The organization of the music files will be explained in a later chapter.
The specific computer used in my music studio came for the ASUS company with an
Intel i5-9400 CPU. The computer uses a solid-state disc drive for all software program
locations while an internal magnetic disc drive is used for all working files. The external
storage uses USB magnetic drives.

4.3.2 Analog to Digital & Digital to Analog Converters

Since the music from the records is analog in nature a device is needed to convert from
the analog to the digital domain for the software to operate. Your hearing is analog so you
must convert back to the analog world to hear the results of your music work. Many of today’s
computers contain a set of chips on the main board to perform these duties. Another method
is to have a converter installed in a PCI expansion slot for the conversions. These separate
PCI plug-in modules are called Sound Cards. Another method uses an external converter
powered by a USB connection. The results of the data conversion are sent in real time to the
computer over the same USB connection used to power the device.
These converters are critical to the success of your music restoration work. Refer to the
Reference Section for detailed information regarding the testing and use of these converters.
 27

For a computer that contain a PCI converter or integrated circuits on the main board the audio
input and output signals should use the line-in connectors. The standard computer colors are
blue for line-in and green for line-out. The definitions of the signal levels for line-in and out
are described in the Reference Section of the Handbook.
If the sound card has a microphone input, it will have a pink color. The microphone
input has an additional gain stage from that of the line-in stage and is not useful for our audio
restoration work. If your computer does not have line-in and line-out connections, then a
converter must be purchased.
One of the converters in my music studio is an external USB powered device (figure
4-4). This converter is from the Focusrite Company and is a Scarlett 2i2 Gen3. The ASUS
computer contains a Realtek Chipset that can also be used as an A to D and D to A Converter.
I use multiple converters in my studio for convenience when using other analog sources than
a turntable.
The operating system software will allow you to vary the gain of the audio conversion
for the line-input and line-output levels via software slider controls. The best result for music
restoration will occur when the software sliders are all set for maximum gain for both input
and output and the use of external gain controls (with real knobs) to adjust the overall
recording and playback volume.
The sample rate and digital word length in bits should be adjustable for the converters
with a maximum rate of 96 thousand samples per second and 16 bits of word length. The
adjustment of these values will be covered later in this section and in the transcribing music
section 7.0 in detail. There are several locations in your music system to change these values
and you must exercise caution that you have really made a change and are using the correct
values when you record your music.

4.3.3 Listen while Recording

The ability to hear the sound that you are recording during the transcription process is
necessary to avoid a poor-quality result. You can achieve this using two separate methods.
First, the signal from the preamplifier can be connected to both the line-in connection on the
converter and to an audio amplifier and speaker. Some preamplifiers have multiple outputs
that enable this signal connection. The second method is to use the converters software to
perform this ability. The software may refer to this setting as loop-back or listen. Often these
settings are in the control panel audio section in Windows software or with the manufacture’s
converter software.
 28

4.3.4 Turntable & Cartridge

Record turntables to playback the music have been made since the beginning of
recorded music. For your restoration work, a used or new turntable can be used. The basic
requirements are:

1. Speed Steps from 33⅓ to 78 RPM.

2. Low Mass Tonearm.
3. Adjustable Skate and Stylus Tracking Settings.
4. Ability to mount and use today’s Stereo Cartridges.
5. Several different Stylus sizes for the Record types.

While used turntables are available, always purchase a model that is working well and
doesn’t need new belts and drive wheels as many of the replacement parts are not widely
available.
A current manufactured stereo cartridge should be used; for example, the Ortofon
series, Grado Labs and others that accept various stylus sizes.
The new turntable models range from simple USB types with a one size fits all to state-
of-the-art models that an audiophile would enjoy. My first turntable was a used Garrard
Turntable and currently I use a RELOOP RP4000 MK2 as shown in figure 4-2.
The USB type turntables that include the A to D converter should be carefully verified
that they can provide the needed sample rate, bit depth, and low noise level for your
restoration work along with an appropriate preamplifier for Non-RIAA Records (see next
section).

4.3.5 Preamplifier and Gain Control

The electrical connections from the turntable to the analog to digital converter pass
through a preamplifier. The preamplifier should have a flat frequency response meaning that
the gain will be the same for all audio frequencies that it amplifies (minimum 20-20,000Hz).
Many record preamplifiers in the past and even those currently available have a frequency
response that is not flat; rather it conforms to a RIAA (Record Industry Association of
America) equalization standard. This RIAA equalization is covered in the Equalization
Chapter and the need for this EQ is explained in detail in the Reference Section. Some
preamplifiers that have an RIAA response also include a means to disable this feature and can
 29

be used. Although the DCart10 Software includes methods to remove the RIAA response
from the transcription, the best approach is to not use the curve in the first place. The RIAA
EQ (equalization) curve is correct solely for LPs (Long Playing Vinyl Records) and 45s and
was not correct for the early records that were made. Very good restoration results can be
obtained by removing the RIAA curve in software while recording non-RIAA EQ records,
but the best method is to start all transcribing operations with a flat preamplifier.
The electrical load for the turntable cartridge produces a significant effect on its
performance to provide a faithful version of the music. Since this is important the manufacture
of the cartridge specifies the correct load on each channel in terms of resistance (Ohms) and
capacitance (Pico Farads). It is the requirement of the preamplifier to have the correct value
as a load for each channel. In most cases the value for the resistance is 47 thousand Ohms and
a very small value for the capacitance. You should verify that the correct load is in place in
the preamplifier by checking the owner’s manuals for the preamplifier and cartridge.
All recordings will use the stereo channels for records whether monaural or stereo. The
correct monaural music sound will be later created in software and in no case should the stereo
channels be wired to achieve a monaural sound. Separate, individual left and right channels
should enter the converter for all types of records. A Diamond Cut Productions Flat
Preamplifier DCP-47K-F (figure 4-3) is used in my studio.
The gain of the recording path must be adjustable and have a maximum value more
than adequate to saturate the Converter’s input while playing any type of record. The best
gain controls are physical knobs that are part of the preamplifier and rotated to change the
amount for each channel. While it is possible to vary the gain with a slider software control
that is mouse driven the use of a physical knob is the best way to maintain the optimum gain
during recording.
If individual gain controls for Left and Right Channels are not possible in your
equipment at this time, a method will be shown in the transcribing chapter to modify each
channel gain in software.
As I was setting up my Music Studio, I found that the Flat preamplifiers available at
the time often had a fixed gain. Since I wanted to have adjustable gain in the recording path,
I used an analog mixer with adjustable channel gain for Left and Right along with an over-all
gain adjustment for the electrical path from Turntable to Converter. One method uses a
Diamond Cut Production preamplifier (DCP-47K-F) connected to the turntable (figure 4-2),
with the output signal of the preamplifier connected to a Behringer analog mixer model Xenyx
502 (figure 4-3). The output from the mixer is then connected to the line-in Realtek Chip Set
in the Computer.
 30

A close look at figure 4-4 shows that the Focusrite converter has individual Left and
Right channel gain controls. A second method is a direct connection from Phono Preamp to
Focusrite converter (the Focusrite converter has more than enough gain to saturate the A to
D).

4.3.6 Playback of Music during the Restoration Process

While restoring your music, you want to hear the results on a good quality audio system
with the speaker’s location near the computer display.
During the music restoration process, you will be listening and watching the display as
the music is played. While you are watching the display of the music you will be hearing the
sound at the same time. This combination of seeing and hearing is very powerful for finding
the location of noise events in the music. The best location for the speakers will be on both
sides of the display and close to your ears. If the speakers are located away from your hearing,
then you may miss a noise burst in the music. Refer to figure 4-1 for a set-up.
The quality of the audio system should be very good so that all the frequencies in the
music can be easily heard. The need for a good playback system, can be a problem since large
physical speakers are needed to reproduce very low notes. A good compromise can be found
with bookshelf size speakers as shown in figure 4-1. Another method for music playback is
to use a set of quality headphones. The front cover of this Handbook shows my first music
studio where I placed monitor speakers on a simple wooden desk.
The playback amplifier in my Audio Studio is shown in figure 4-5. The power amplifier
is located behind the monitor and the preamplifier is on top of the speaker on the right side.
When the noise from the music has been removed and the music file is ready for the
final effects to be use, the type of playback system has another important function. The
quality of the restored music is greatly influenced by the type of playback equipment used.
When you play your restored music, you may use a variety of devices from headphones, high
end audio systems, and portable CD players. A restored song should sound good on as many
types of audio systems as possible.
Today’s production of commercial music uses special monitor speakers that simulate
the most common type of listening that will be used. This extra level of expense maybe useful
for some however you can use your own judgement to produce good results by playing the
restored music at home or in your car.
 31

4.3.7 Setting Windows Privacy for Microphone

The Windows Operating System contains a Privacy option from the main Settings
control that controls what software programs that can communicate with the record function
on your Analog to Digital converter. Note that the Windows Software uses the term
Microphone rather than line-in or A to D.
To have the Diamond Cut software program control the A to D converter, the
Microphone setting in the Privacy control for Desktop Apps must be turned On. Refer to
figure 4-6 for the correct settings:

Figure 4-6 Application Control of A to D

Note: If you have not used Diamond Cut Software in the past then you will not see the Icon
in the history under name. You will see the Icon when used.
 32

4.3.8 Control of Sample Rate for Converters Internal & External

The DCart10 software provides a user interface to the sample rate and number of bits
for the file that you will be creating when the record button is selected.
Figure 4-7 shows the settings for sample rate and number of bits that you want for the
music file. The actual control of the converter will be via Windows Drivers or specific
software from the converters manufacture. Figure 4-7 will show what has been selected and
what the header file for the recorded music indicates (later steps will synchronize theses’
settings and the converters values).

Figure 4-7 Sample Rate and Bit Resolution Settings

The operation of this part of the DCart10 software requires that the software commands
in DCart10 are in-turn received and implemented by the various Windows Drivers and any
User added Converters. There are times when this does not happen. The file that was created
by DCart10 software will have a header that contains the information selected in figure 4-7
but unless the specific converter settings match up, you will not record the desired values.
Let’s see why this can happen.
 33

The Windows Operating system controls the sound system properties via a software
driver. One path to this driver is through the Control Panel Hardware and Sound window,
and then Sound windows’ option. At this point, the various tabs can direct you to the recording
and playback sections which in-turn will allow control of the sample rate and bit resolution
though the advance tab. The control of the A to D and D to A values at this point may or may
not be what was set in Figure 4-7. One requirement to enable DCart10 control of the converter
is the use of a software option in the recording tab called exclusive control. This box must be
enabled so that the DCart10 software has total control over the settings. This exclusive control
option may still not work at times due to the Windows Drivers operation. In addition, external
USB converters will at times use non-Windows Audio drivers to control their product to
avoiding Windows Drivers. Often these converters will have a unique driver for DAWS
(Digital Audio Workshops) to correctly function but when they use a Microsoft driver, they
cannot always communicate with each other.
If you want to use an external converter you will need to check with the manufacture
of these devices so that you can set the sample and bit values correctly.
This is a very important part of your restoration process and will be covered again in
more detail in the Transcription Chapter.

4.4 Overview of DCart10 Software Operation for Restoration

The Diamond Cut Productions Company provides several useful sources to learn how
to use their product. The Getting Started Guide and User’s Guide should be reviewed to learn
about the operation of the software. An overall description of the restoration operation will
now be described, and addition chapters will add detailed information.

4.4.1 The Operation of the Software

After the Diamond Cut Productions DCart10 software is installed on your system for
the first time, you will have several options to pick for the look and feel of the product. A
screen shot of the initial layout that I use is in figure 4-8:
 34

Figure 4-8 Starting Screen

At this point the various preferences have been set under the edit tab, the toolbars that
you want are displayed, the amplitude graphs are show on the right, and you have selected
Classic Mode for editing. Next, a source will be opened with a typical result in figure 4-9:

Figure 4-9 Source File Opened

 35

The fundamental operation of the process is to start with a source file containing your
music to restore and then by using a series of filters and effects remove the distortions in the
music.
The result of the software changes to the source file that uses filters or effects will be
located in the destination file. If you are using the classic edit mode, then you will have the
original source file at the top and the destination file directly below the file. The toolbars in
the top of the window allow you to play the selected source or destination file. The result of
applying any operation to the source file will, at most times, produce a destination file. The
result of a filtering operation on this source file is shown in figure 4-10:

Figure 4-10 Source with Filtered Result in Destination

Notice in figure 4-10 the original source file has two channels while the resulting
destination has one channel. The destination file is highlighted to indicate that it is now the
active window and the multifilter used to create the destination is shown on the destination
window.
The documentation for the various filters and effects will show you how to use each
one of these software tools and as you use them you will learn what they do and how they
work.
Some of the operations that you perform on the source file will only modify the file
without creating a destination file. For example, if you want to remove a section of the source
file, then you will highlight the section and cut it out by pressing the control and X key at the
 36

same time. A new destination file will not be created; instead, the source file will be changed
in length. You can recover the removed section via the undo edit feature if you removed the
wrong part.
The DCart10 software has many features that will become second nature to you after
you understand what they do. The software does contain help sections and the reason that the
software looks complicated is due to the many options allowed and not due to what you are
trying to do to the music.

4.4.2 Multifilters

The normal restoration process will involve several individual applications of filters
and effects to transform the original music file into a final clean sounding piece of music.
Each application of a software algorithm upon the source file requires data conversions and a
very useful technique is to place several operations together in a user defined multifilter so
that a major operation on the source file can be implemented in one custom filter that you
create. By having the individual algorithms connected in a large multifilter the number of data
conversions and potential loss of accuracy will be minimized. This multifilter concept will be
used in this Handbook for all the restoration operations.

4.4.3 Music Storage and CD Creation

Within the DCart10 software, you can create a music library database that stores all
fully restored files along with a very useful search and playlist creation. The software also
provides a method to create specific CDs that contains your music in this Tune Library. The
Tune Library will be covered in detail in the Advance Enhancement Section.
 37

5.0 Music File Organization

How you organize your music files on your computer is important and is the first step
in the restoration of your music. You will be able to save a large amount of time during the
restoration process by creating a specific method for organization of the produced music files.
The reason that the organization of your files is important is that you will often find
new methods for optimization to various filtering and enhancement operations after you have
spent considerable work cleaning up the noise from the music. If the resultant intermediate
work is retained, then you can return and build upon your past work while using your new
techniques rather than starting at the beginning again of your restoration work.
Another reason to organization your music files is that you may want to return to the
original transcriptions if you made a major mistake by erasing a file or changing the music in
a manner that cannot be un-done.
Today’s digital storage devices provide the ability to store a large amount of music
files for a low cost. The ability to provide multiple versions of the same song in different
stages of restoration is quite reasonable and very practical.
A consistent method that you use for placing the music in file folders will not only help
you find a specific piece of music on your storage device but will also provide a predictable
method to easily bring your music into a music library. The DCart10 software provides a
useful music tool referred to as the DC Tune Library; and the structure of the music folders
will work with this library. A later chapter in the Advanced Section of the Handbook will
detail information about the DC Tune Library.
The music files are in three major locations which are:

1. Original Recordings. This is where the music’s transfer from the A/D converter is
stored. These recordings will have had no equalization applied to them (flat) and they
have been played at the correct speed. Note: The initial raw recording may need some
software processing before they are stored as an original file. This processing will be
discussed in the chapter on transcribing the music.
2. Restoration Files. This is the working location for the music restoration process.
Various sorting of songs and levels of restoration are located here.
3. Restored Music. This is the location for the latest versions of the restoration results that
can be used for listening or for making CDs (or other final versions).
 38

5.1 Music File Type Used in The Handbook

The wav file type is used in the Handbook. This is the best to use as it’s a universal
music file and is not compressed. While the size of these files can be large the need for large
storage space is not a problem with current storage technology. The wav file is the type that
the DCart10 software operates on and is a standard within Microsoft Software.

5.2 General Organization Method

The folder names for the locations are shown in an italic style type.

Music

The top level on your mass storage device will start with Music as that is what we are
storing.

Original Source Music Recordings

This file folder is below the Music top level and contains the Original Music
Recordings. It’s very important to have a special place to keep the initial transfers of your
music from the records.

Restoration Files

This file folder is below the Music folder and is the master location for all your working
restoration files. This is the place where you perform your restoration. Various levels of
restoration will be created so that if you want to return to a previous level of work for possible
improvement you can do so without having to re-do all of what you have previously done.
This concept is very important and was learned the hard way!
The layout of the restoration files contains various Level folders that correspond to the
place in the restoration process that the music files presently reside. For example, Level 1
contains the original music files that are ready for multifilter noise removal. Level 1A contains
the music files that are so noisy that new technology will be needed to fix them. The Level 2
files have had the noise removed and proper EQ (Equalization) applied.
 39

The layout of the Level folders within the Restoration Files will have additional
structure to separate the song titles and artists. The specifics of the layout will be shown later
in this chapter, while the details of how to do the restoration work will be described in later
chapters of the Handbook.

Restored Music

This folder is below the main Music folder and contains the final product of your
restoration work. These music files may come directly from Level 2 in the Restoration Files
Folder or may have some additional enhancement applied to the music.

5.3 Specific Folder Naming within Major Locations

Under each major location (Original, Restoration Files, and Restored Music) will be a
similar layout for the music files with changes needed to accommodate the difference between
individual records, albums, and levels of work. While the original concept for a record Album
started with paper folders containing 78 records (Album of Music), the use of the phrase
Album has become associated with LP records, since LP records often contain multiple
songs. The LP Albums can also contain songs from either one artist or a theme of music.
Due to the differences between Single Records and Albums, a slightly different folder
arrangement for each will be used. The DC Tune Library can work (with slight preference
changes) with these different layouts.

5.3.1 Original Source Music Recordings Layout

Individual Records

This layout is used for any single recording of a song within the Major Folder Original
Source Music Recordings. The first folder is the record type such as 78 Electric or 78 Acoustic
(these terms will be explained in the following chapters and in the Reference section). The
next folder down will be the Artists Name followed by the title of the song. An example for a
78 RPM Song is shown with a backslash (\) separating the folder names:

Music\Original Music\78 Electric Records\Al Dexter\Pistol Packing Mama.wav

 40

LP Albums

This layout is used for any LP record within the Major Folder Original Source Music
Recordings. The first folder is the Artists Name followed by a folder with the Title of the
Album. Next folder down will be the song in this Album. An example using the same symbols
as previous used:

Music\Original Music\LP\Alicia Bridges\Disco Round\I Love the Night Life.wav

5.3.2 Restoration Files

Additional folders are used to define the stages of our music restoration work. The *
character will indicate that the number used will be 1,1A, or 2 for the different levels of work.
These levels contain your restoration results and music with the same artist and title will be
in both Level 1 and Level 2. The level method will be explained in more detail in the chapter
on General Noise Removal.

Individual Records

This layout is used for any single recording of a song within the Major Folder
Restoration Files. The first folder will be the record type such as 78 Electric or 78 Acoustic.
The next folder down will be the Level Number followed by the Artists name followed by the
title of the song. This layout will be used in this example with a backslash (\) separating the
folder names:

Music\Restoration Files\78ElectricRecords\Level *\Al Dexter\Pistol Packing Mama.wav

LP Albums

This layout will be used for any LP record within the Major Folder Restoration Files.
The first folder will be the name LP. The next folder down will be the Level Number followed
by the Artist Name. Next down will be the Title of the Album followed by the song in the
Album.
 41

An example using the previous used symbols:

Music\Restoration Files\LP\Level*\Alicia Bridges\Disco Round\I Love the Night Life.wav

5.3.3 Restored Music Files

This folder will be used by the DC Tune Library for your restored music.

Individual Records

An example is:

Music\Restored Music Files\78 Electric Records\Al Dexter\Pistol Packing Mama.wav

LP Albums

An example is:

Music\Restoration Music\LP\Alicia Bridges\Disco Round\I Love the Night Life.wav

5.4 Summary

A structured system of file locations for restoration work will provide access to a
specific song by an artist and return to your restoration work. Also, new methods of
restoration can be applied without a re-do of previous work.
The amount of time to restore a song can be quite long. You will want to build on your
work and not repeat what you have already done. If you find that a restoration result damaged
the song you can always return to the original recording and start over by using a structured
file system.
 42

6.0 Music Equalization & Removal

6.1 Overview

The reference section describes how audio technology began by using the sound power
of the music to directly record the sound waves and then progressed to the use of electrical
devices to amplify the music collected by the microphone before it was placed into the master
recording. The time during which only the sound power made the recording is referred to as
the Acoustical Era or Acoustical Recording and for electrical recording the word Electric will
be used. LP and 45 RPM will not need any extra labels as these recordings only used electrical
devices.
Electric recording of music modified the music before making the master recording for
two important reasons. The first reason was that modification to the low music frequencies
was needed to avoid a very wide groove width to contain the amplitude movements. The
second reason helped to reduce the surface noise of the record allowing the higher musical
frequencies to be heard. With these special modifications made to the frequency of the music
the low frequencies could use the same groove width as the medium and high frequencies and
the surface noise of the record would be reduced.
The application of frequency modifications prior to recording the master record and
the application of the inverse of these modifications after playback of the record produces an
accurate sound of what was heard by the recording engineer in the studio. Details will follow
in this chapter with additional information found in the Reference Section.

6.2 Definition of Terms

1. Acoustical: Only sound energy was used to produce the master recording. This was the
original method to record music.
2. Flat: Means that the amplitude of the playback music that passes through the electrical
circuits to the A to D converter are not modified depending on their frequency value.
3. Equalization or EQ: A modification is made to the signals amplitude as it passes
through the circuits to a specified frequency response curve (often to counteract the
effects of the original electrical recording).
4. Constant-Amplitude refers to a modification of sound waves so that the amplitude of
the signal will remain constant as the frequency decreases (for constant input signal
 43

power). To achieve this, the signal power of the actual recorded sound wave must be
reduced in relationship to the frequency change.
5. Constant-Velocity refers to a property of sound waves such that the maximum velocity
of the signal stays fixed as the frequency changes (for the same value of signal power).
The amplitude of the signal must change in relationship to a change in frequency to
maintain a constant-velocity and will steadily decreases with frequency (for the same
value of signal power).
6. Turnover Value: A frequency where the response of the sound wave changes from
Constant-Velocity to Constant-Amplitude.
7. Pre-Emphasis: A time constant in the filter’s circuit (which is a specific frequency)
where the Constant-Velocity Response changes to an increasing velocity with
increasing frequency (Acceleration).
8. Rolloff: The amount of amplitude correction needed to remove the added boost from
the Pre-Emphasis application.
9. Reference Point: A frequency that had no modification made to its amplitude; used for
dB (decibels) measurements which requires a reference or 0 dB value.

The Tradition way to refer to the modifications are:

1. The modification to the playback signal uses the Curve Name.

2. The modification to the recording signal uses the phrase “Reverse” before the Curve
Name.

For example, the words Reverse RIAA Curve describe what is used to record the initial
master record. The applied RIAA Curve would then remove the original record modification.
While this usage may seem backwards it is the method used in the music business.
In the chapter on transcribing records, it is recommended that for the transfer of the
music from the record to the digital domain a flat preamp should be used which means that
no equalization will be applied by the preamp circuits.
If a flat preamplifier was not used, specific methods will be shown to produce a flat
recording from an initial EQ applied by the preamplifier. Since the use of a flat preamplifier
means that no correction of equalization is applied; when the music is heard using this type
of preamplifier from an electric recorded record the sound will be quite different when the
correct EQ is used. The electric recorded music with no EQ (flat preamplifier) will sound
tinny and thin to you and that is O.K. at this time. If the preamplifier applies EQ when an
 44

Acoustic record is played, rumble and lack of high frequencies will be heard. If the Acoustic
record is played with a flat preamplifier the music will have a natural sound.
When you listen to the music during the restoration process you will want the proper
equalization applied so that you can hear the music in a natural manner. The reason for the
flat recording for the original recording is related to the operation of the noise filters and to
give you flexibility to apply the specific EQ that you will use to create a restored music file.
Several different methods to apply and remove the frequency modifications in software
are shown.

6.3 DCart10 Equalization Tools

The DCart10 software provides two methods to apply a frequency modification to the
recorded music to achieve the proper EQ or to remove specific types of EQ from the
recording. The Paragraphic EQ software has a large selection of different frequency responses
to apply to the music along with a clever graph that displays the response. Another method
uses the Virtual Phono Preamp that has options that you can select with software buttons.
Both methods have several preset responses available (as do many DCart10 devices). Refer
to Figure 6-1:

Figure 6-1 Software EQ Methods

Let’s look further into these two different methods; first the Paragraphic EQ with a 500
Hz turnover is shown in figure 6-2:
 45

Figure 6-2 Turnover at 500 Hz

Notice how the low frequency response changes at the turnover frequency to have a
gain increase of 6 dB per octave (doubling of frequency) and that this gain continues to
increase with lower frequencies. Prior to the turnover value, the gain versus frequency was a
flat response. The reference point for the turnover frequency is 1000 Hz and at 500 Hz the
response is +3dB from the 1000 Hz value. This reference value of 1000 Hz is often assumed
and not stated.
There is an additional option on this filter namely the Low Freq Shelf option. What this
option does is to take the normal 6db per octave slope of the curve and changes its slope to a
flat shelf around 150 Hz. This change can help to reduce the amount of low frequency boost
to the recording at the lowest part of the music frequencies where rumble and other sources
of noise could be present. Refer to figure 6-3 to see the effect this option had on the same 500
Hz turnover curve.
 46

Figure 6-3 Turnover at 500 Hz with Low Freq. Shelf

Another method for adding or removing EQ is the Virtual Phono Preamp. This
multipurpose software algorithm may reduce steps in the restoration by combining many
functions in one device. Refer to figure 6-4:

Figure 6-4 Virtual Phono Preamp 78 Record

 47

Notice that the selected settings indicate that the original recording was from a flat
preamp and the desired playback curve is for an American 78. These setting applies the same
500 Hz turnover setting as discussed before with an additional filter selected (30 Hz) to
remove rumble. In addition to the turnover response, this filter allows you to modify each
channels volume, observe the amplitude balance and change the tone (shelf tone controls).
The comments American and European are not to be used as a universal setting for
either 78 records made in USA or Europe. Many records made before World War Two used
a turnover of 250 Hz and this value is labeled European in this filter. The American setting
uses a turnover of 500 Hz and was common for records made after WW2.

6.4 How to Produce a Flat Recording from a Non-Flat Preamplifier

When the first electric record players were introduced, the controls were very basic and
consisted of a volume control and a simple turnover circuit. As time went on the audio
equipment became more and more complex and the record companies started to develop
unique equalization curves for their own records. This led to preamplifiers with numerous
switches to allow you to set the EQ values. The audio equipment companies, and the record
companies finally settled on a RIAA EQ that was standard for all records and this EQ became
built-in to many of the preamplifier’s phonograph input circuits. Today these older
preamplifiers with selectable EQ values are not common and preamplifiers with just the
RIAA curve are the ones likely to be available.

6.4.1 Two Types of RIAA EQ Filters are Present, Which One to Use?

The DCart10 software contains two filters that can remove the RIAA curve from the
preamplifier. The Paragraphic EQ filter was the original method used to apply different types
of EQ curves while the Virtual Phono Preamp was a later software development. The Virtual
Phono Preamp follows the various EQ curves to a higher precision than the Paragraphic EQ
software option and for this reason this filter should always be used to remove the RIAA curve
from a Preamplifier (for those who do not have a flat amplifier). The Virtual Phono Preamp
uses the phrase Acoustical which can also mean Flat in the following example. To produce a
flat (no EQ) recording from a RIAA Preamplifier when playing a 78 Electric Record you
would set the software settings to those in figure 6-5:
 48

Figure 6-5 RIAA to Flat Recording

In this example the acoustical setting gives us a flat recording for the restoration work
by incorrectly telling the software what type of record we have. The record type is really an
American 78 RPM record but by using the acoustical option, the Preamp RIAA curve will be
removed. The new flat recorded file is then saved for further processing as the Preset title
indicates.
The Paragraphic EQ filter contains a preset to remove the RIAA curve but performance
is better using the Virtual Phono Preamp.

6.4.2 The Types of EQ that can be Removed

The DCart10 software has many options to apply EQ to the music waveform but the
software can only remove two types of standard equalization curves. These two types are the
RIAA EQ and the NAB (National Association of Broadcasters) Tape curves. The RIAA curve
can be removed using Virtual Phono Preamp software and the NAB Tape curve can be
removed with the Paragraphic EQ software. Because the DCart10 software has this limitation
on the number of EQ curves it can remove, the various noise removal methods were
developed to apply the EQ in specific stages.

6.5 How to Determine if the Record was Acoustical or Electrically Recorded

 49

Knowing the type of recording that was used for the record is necessary for music
restoration. The choices are either Acoustical or Electrical recording; Acoustical recording is
a mechanical operation where the sound energy moves a recording cutting head in
synchronism with the music waves to store the music on a master record verses Electrical
recording that captures the music with a microphone and after amplification and modification
provides power to the groove cutter to place the sound into the master recording.
One method to determine the type of recording is to listen to the music as it is played
back. This method uses the source recorded from a flat preamplifier while you toggle the
virtual phono preamp on and off as you listen to the song. The Record Type Option should
be selected for what you think the record type is. The use of the preview control and bypass
makes it easy to apply and remove the EQ while listening to the song. For an example, if you
have the flat recorded file as the source, bring up the phono preamp, select preview and play
the music as you toggle the bypass option. Refer to the figure 6-6:

Figure 6-6 Checking the Music for Equalization

 50

In this example turning the bypass on and off during a Preview makes a difference that
is easy to hear and since the music sounds correct with the Virtual Phono Preamp on, you
know that the song was electrically recorded. The opposite condition with the sound changing
slightly or not at all means that the song was originally acoustically recorded.
After hearing a number of songs, the difference between the Acoustic recordings and
the Electrical recordings will become second nature. If you have a questionable song that
could be either, there is another method that uses the software’s spectrum analyzer’s ability
to provide a frequency profile of the music to help you decide between the two.
The use of the spectrum analyzer can show the type of recording that was used since
the acoustic and electrical recording each have unique frequency responses. In the case of the
Edison Diamond Disc records this method is often needed since the quality of the Edison
acoustic records was state of the art at the time and the records have a very nice sound for
acoustical records that sometimes sound like an electric recording.
While the use of the spectrum analyzer while playing material from a flat transcription
will often show differences between electrical and acoustic recorded records; the difference
is somewhat easier to see by using the American 78 setting on both types of records along
with the use of the Spectrum Analyzer. Let’s look at an example using arbitrary 78 Electric
and 78 Acoustic records. Both music selections were recorded with a flat preamplifier and
then had the American 78 setting applied with the software Virtual Phone Preamp while the
spectrum analyzer was running. The specific American 78 setting may not be the best for
this specific electric recording, but the method shown here works using this type of setting.
The acoustical recorded spectrum analyzer picture (with the American 78 setting on)
is shown in figure 6-7:
51

Figure 6-7 Acoustical Record with American 78 EQ applied

Next, the Electrical recorded spectrum analyzer picture is shown in figure 6-8:

Figure 6-8 Electrical Recording with American 78 EQ

 52

In both cases, the equalization (American 78) applied a boost to the lower frequencies
(turnover 500 Hz). This boost works only if there is music energy present in this low
frequency area to increase. In the case of the acoustical recorded music the spectrum below
200 Hz falls off to indicate that there isn’t any music present to boost. This lack of lower
frequencies contrasts with the electrical recording in figure 6-8. Another area to look at in the
frequency spectrum is the higher frequencies for a contrast between the two recording
methods. Figure 6-7 shows a sharp drop-off over 3 KHz whereas figure 6-8 shows energy
present out to 10 KHz.

6.6 The Correct Equalization for a 78 Electric RPM Record

The best approach for selecting a value to compensate for the frequency modifications
is to use the published tables in the DCart10 software help files. Refer to table 6-1:

Table 6-1 78 Turnover Values from the Help File

 53

The text in the help file states this equalization can be applied using either the Virtual
Phono Preamp or the Paragraphic EQ options; however, the Virtual Phono Preamp will apply
a more accurate result but is limited to two 78 RPM setting ; American with a turnover of 500
Hz or European with a turnover of 250 Hz. The Paragraphic EQ has a wider range of values
to choose from.
This table provide a useful starting point but the record producers have, since the
beginning of recorded sound, imposed their own biases on the sound of the music. What this
means is that as you restore the music you can modify the equalization that you apply to shape
the sound for what you believe to be correct. The modification that you make to the published
EQ numbers will be made to the music after the Level One operation (details in the noise
removal chapters). In this manner, you can always return to the original sound and make
different types of modifications without having to start at the very beginning with an original
recording. The only EQ curves that can be removed are the RIAA and the NAB Tape curves.
You cannot remove a 78 RPM curve once you have applied it to a flat recording.
This lack of standardization by the record companies reinforces the reason why you
want to start with a flat recording. When the starting point is a flat recording, the conversion
to a specific type can always be changed later when you learn more about the recording
process on a specific record since the original has no EQ applied.
This example will clarify the use of equalization. You start with a recording of a
popular swing tune from the late 1930’s. The record was made by the Victor Company, so
you have some choices from Table 6-1 of a turnover value of 200 Hz or 500 Hz. The first
application of EQ uses the American 78 RPM (500 Hz Turnover) setting. After the file has
had the noise removed the music file is placed in the Level Two music folder using this EQ.
Up to this point you have chosen to use EQ settings that have been recommended to you
independent of knowing what was specifically done during the mastering of that record. The
next step in the restoration is to use your hearing to modify, if needed, further changes to the
records sound. If you find that a turnover value of 200 Hz sounds better than the 500 Hz value
then you can return to the non-EQ file (Level 1) and easily produce the better sounding file
(Level 2) by changing the EQ in the Level 2 production and then use it for your final restored
music.
There is no reason that your restoration work must adhere to a strict method of using
the recommended EQ settings. As your ability to restored music improves, you can always
return to the starting point (Level One) and add to your previous work.
These terms Level 1 and Level 2 will be explained in more detail in the noise removal
chapters.
 54

6.7 Note Regarding Pre-Emphasis on 78 Electric Records

The original technical work performed by the Western Electric Lab in 1925 suggested
that an improvement in the recording of high music frequencies was possible by boosting
their amplitudes above a certain frequency during the recording process and in turn followed
by an inverse operation during playback. Additional information can be found in the reference
section. While this technique was known to the early music recording companies, widespread
uniform implementation would have to wait till the adoption of the RIAA curves. However,
some of the record companies did implement this method and the literature in audio books
and technical articles confirms that it was possible for a record to have its high frequencies
increased from the original recording. How can we deal with this unknown high frequency
boost?
The location for the restoration work of modifying the upper frequencies can best be
performed later in the enhancement chapters and therefore the EQ for 78 RPM electric
recorded music should stay as just a turnover curve for the music up to Level 2 without adding
a roll-off value.

6.8 Special EQ for Edison Electric Recordings

The history of the Edison Phonograph Company showcases an organization that was
independent from the rest of the music industry in all matters. Edison used a disc groove that
had up and down or hill and dale variations for the music. The Victor company and many
others used a back and forth or lateral variations for the music. While Edison’s original patent
for the phonograph claimed the use of both lateral cut and vertical motion to capture the
sound, he believed that the vertical method gave a better acoustical performance than did the
lateral method. Edison’s many patents helped to create the invention of the vacuum triode
(Edison Effect) and the microphone, yet he did not adopt the new electric recording method
(1925) until much later than his competitors (1927). Very little written information is
available regarding the methods that were used by the Edison Company for the electrical
recordings at this time. From the discussion on the need for equalization recall that the stylus
motion for the Edison Diamond Disc record is a vertical motion and while the amplitude does
increase with low frequencies, the potential swing into an adjacent groove does not have the
same limitation as with a horizontal motion.
The Edison Electric recording improved the frequency response of the record believing
that most playback would be via acoustic or mechanical means. What this means is that some
 55

of the low frequencies would not be reproduced by the record player so that the low end could
contain significant rumble and noise that would be present in the record but not heard using
the Edison record player. The use of a standard 78 RPM turnover curve with the Edison
Diamond Disc Electric records will often yield a noisy and un-natural sound for the low
frequencies. The low frequencies can be boosted and will improve the overall sound, but the
method requires extra steps. Specific information will be shown in the enhancement chapter
for Edison Electric Diamond Discs. For the Diamond Disc records that were electrically
recorded the best result is to not use any specific bass boost or turnover value during the Level
Two or Basic restoration work.
For the Edison Needle Type lateral records, the table in the Diamond Cut help file
(Table 6-1) shows a turnover value of 500 Hz. These records used a lateral motion for
recording sound and were produced for a very limited time.

6.9 EQ for LP and 45 Records

In a previous section (6.4.1) the virtual phono preamp was used to remove any EQ from
a preamplifier as it implemented the RIAA curve better than the Paragraphic Filter. When
LP or 45 RPM records need to have the EQ applied, either filter can be used. The reason is
that these records can lose some high frequencies due to groove wear and the Paragraphic
Filter may provide an improved sound because of a slight difference in high frequency
implementation.
The chapter on the details of the LP and 45 RPM record noise removal will use RIAA
EQ from either the Paragraphic or the Virtual Phono Preamplifier software depending on the
condition of the record.
 56

7.0 Transcribing Records

The process of taking the recorded music from the source and converting it to a digital
file is known as transcribing a record and starts with converting a recording to a raw music
file.
Next, any corrections to the raw recording to correct the speed, make it a flat recording,
and balance the channel gains are made. Lastly, trim away any extra recording during the
beginning or end of the music with the result an Original File.

7.1 Definition of Terms

1. Raw Recording. The phrase raw refers to the first conversion of the analog signal to a
digital value. No additional signal modifications have been performed on the resulting
digital file. The recording will contain the entire result of the Analog to Digital
conversion from start to finish.
2. Flat Recording. Any EQ applied to the Raw Recording has been removed.
3. Balanced Channels. The phrase balanced means that the individual gains of the Left
and the Right channels are set to produce equal recorded Left and Right Channel
Amplitudes.
4. Trimmed Recording. The phrase trimmed means that the sections of the digital
conversion from the analog recording that occurred when the stylus was not in the
record groove have been removed.
5. Original Recording. The phrase original means that the recording has had the results
of terms 1,2,3 and 4 applied to the digital file.

7.2 Equipment Set-up

The tools chapter described the audio equipment needed for the recording and playback
process. Refer to figures 4-1, 4-2, 4-3,4-4, and 4-5 for a possible setup for the transcribing
process. Some of the material will be repeated from chapter 4 with added detail.
Check your privacy setting (4.3.7) to verify that Desktop Applications can use your
microphone.
 57

7.3 Preamplifier and Turntable Speed

In the tools chapter a flat preamplifier with adjustable channel gain control was
specified for transcriptions. The flat preamplifier refers to an amplifier between the turntable
and the analog to digital converter that amplifies all frequencies with the same gain. If you do
not have a flat preamplifier i.e., the one you have was designed to play LP records with the
RIAA curve, then you can still use this type of preamplifier, but you will need to apply an
extra step. This extra step is discussed later in this chapter. This will occur after the raw music
file is made but before the file becomes the original recording. If you do not have individual
channel gain adjustment in hardware, then a software feature within DCart can be used to
change the gain for each channel. This technique will be shown later.
DCart can compensate for speed differences between the correct record speed and the
turntable speed. For example, a 78 record was designed for playback at 78 RPM (revolutions
per minute) whereas a modern turntable will often have a position for 33⅓ RPM or 45 RPM
and not a 78 RPM setting. In this case the record will be played at 45 RPM and using software,
converted to sound as if it was played back at 78 RPM. While there maybe times that a record
benefits from being played back at a slower than designed the best method is to playback the
record at the speed it was recorded at. If the record is played back at a slower speed than the
original, music frequencies will be changed in proportion to the speed change. For example,
let’s say that the 78 RPM record has a strong bass note at 50 Hz. When that same note on the
record was played back at 33⅓ it would be at about 21 Hz! Some of these new low frequencies
could be below the frequency that your preamplifier can respond to and will be lost for your
restoration. The best outcome is to play back the records at the same speed that they were
originally designed for.
The one exception to this rule is for the Edison Diamond Disc records which were
recorded at 80 RPM. The conversion from 78 RPM to 80 RPM is very small and often needed
since 80 RPM turntables are not common.
The selection of the cartridge and stylus for the playback of the records is very
important. The music information is contained in the motion of the grooves on the record and
the job of the stylus and cartridge is to convert this motion into an electrical signal. Therefore,
you want the stylus size to fit the specific record groove.
 58

7.4 Cartridge and Stylus

The selection of the Cartridge and Stylus is very important in order to extract the best sound
from the record. The reference section provides addition information regarding the choice of
Cartridge and Stylus for your transcription.
The Cartridge should be a stereo model that fits your turntable and has the ability to
accept different types of styli. Many people used a Stanton 500.V3 cartridge since the price
was moderate and different types of styli were available. This brand is not currently being
produced although old stock can be found. There are choices available today; Ortofon, Grado,
Nagaoka, and others. When the cartridge is selected the preamplifier should apply the
recommended electrical load.
The stylus size is critical to the quality of the playback of the sound. If the tip rides low
into the groove for a lateral record it will scrape the bottom and add significant noise to the
music. On the other extreme it can ride too high and miss some of the recorded sound in the
groove. Styli are described using a dimension in either mils (thousandth of an inch) or µM
(millionth of a Meter) for the radius of the tip for a simple shape; and two numbers for a
complicated shape. When two numbers are used one is for the larger radius and the other is
for the smaller radius (an elliptical stylus has a major and minor radius).
For Edison Diamond Disc records a stylus has to ride in the bottom of the groove. Most
Diamond Disc records can use a “DJ” type of stylus that has a conical shape and a radius of
0.7 mils. For very worn records, a 3.75 mils conical shape stylus from the Expert Stylus
Company (located in England) works well. The Reloop OM Black 0.7 mil DJ stylus from
Ortofon is used for Diamond Disc records in my Audio Studio.
For the 78 Acoustic and Electrical records, a 2.7 mils stylus (Stanton D5127) has been
used in my Audio Studio. For today’s LPs and 45s the standard size is 0.7 mils x 0.4 mils and
is a good choice. It is possible to spend quite a bit of time and money to obtain a large set of
different styli for the best transcription result. The amount of wear on the groove of the record
can often be avoided by having the stylus ride the groove in a new location that is removed
from the distortion and damaged section of the groove. Trial and error in stylus size can be
used to find this new location.

7.5 Channel Selection

All types of records will be transcribed using a Stereo Cartridge with Left and Right
output channels. Equal preamplifier gain is applied on each channel and all records are
 59

recorded without any frequency equalization or flat. All further signal processing will be
performed in software to the original left and right channels during the noise removal work
to produce the desired outcome and if needed channel balance.
The use of stereo recording for the transcription allows, if needed, the later separation
of the vertical or horizontal components of the stylus motion via software routines. The reason
that this separation can be done is that when the stereo recording method was originally
developed the designers wanted a technique that would be compatible with previous
recordings that had one channel (monaural) of music. Since the previous monaural records
used a horizontal motion the solution to adding a second channel (Stereo), was to change the
stylus horizontal motion to one using two 45-degree vectors with the left channel on one side
and the right channel on the other. This 45-degree motion allowed a new vertical component
to be added to the lateral component resulting in two channels of music, left and right, in the
record groove. The original horizontal (monaural) would come out equally on the left and
right channels allowing the older records to be heard. The older hill and dale technology that
Edison used was merged with the lateral motion to create Stereo recording.
Since the cartridge coils respond to 45-degree vectors, by performing math operations
you can separate either the Vertical or Lateral Motion. The fact that the Vertical or Horizontal
groove motion can be pulled out of the stylus motion means that all types of records can be
played back using a stereo cartridge for the original transcription.

7.6 Sample Rate and Bit Depth Selection

The use of 96 kHz (96 thousand samples per second) is recommended for all
transcriptions. This high sample rate is useful for restoration work since the software
algorithms for noise removal will benefit in two ways; first the higher sample rate allows very
high (ultrasonic) energy to be recorded which will aid in finding impulse noise (which is
largely high frequency) and second the large number of samples per second provides many
pieces of information to the algorithm that restores the missing music section during noise
removal operations. The fact that the noise filters have more samples per second available to
them often allows a more aggressive setting for these filters with a resulting cleaner sounding
music verses a lower setting of 44.1 kHz samples per second.
A sample rate of 44.1 kHz (44.1 thousand samples per second) can be successfully used
if the equipment only supports this sampling rate. The noise removal can be more difficult
and will require more manual noise removal. While this rate has the advantage that the size
 60

of the music files will be less than the 96 kHz, the cost of digital storage is low and should
not be a factor in selecting this sampling rate.
The DCart10 software allows a sample rate conversion via a re-sampling algorithm,
however, it is NOT recommended to use this conversion to a higher rate than the original
recording was transcribed at to achieve better noise removal. If the original rate is 44.1 kHz,
then keep this rate for your restoration work.
The word length of the recording is set to 16 bits of digital depth. This value is more
than adequate to maintain a high signal to noise ratio needed for digital processing. Refer to
the reference section for additional information that justifies the use of 96 kHz sample rate
and a digital word length of 16 bits. The reference section contains additional information on
the selection of the optimum sample rate and bit depth. Some analog to digital converters
provides only one choice of 24 bits for the digital word. This value can be used but the best
method is to select 16 bits by using the options in the converter driver or the control panel in
Microsoft operating system.
In chapter four the control of the sampling rate and bit length was described as needing
both the software drivers for the converter and the DCart10 software to communicate with
each other. This communication will not work at times with some external converters for
DAWS (Digital Audio Workstations) and internal converters that use unique software drivers.
An additional complication is that Microsoft audio drivers will re-sample digital files to
different rates at times to accommodate user needs without the user aware of the re-sample.
It is very important that the desired sample and bit rate is performed to produce the music
file. The next method will ensure that it happens for the music restoration work.

7.6.1 Setting the Converters Sample Rate and Bit Depth for Recording

The first example will use an external USB converter made by Focusrite (Scarlett 2i2
Gen3). Refer to Figure 7-1 using the Focusrite software.
 61

Figure 7-1 External USB (Focusrite) Settings

The Focusrite company provides additional information regarding the operation of

their device with the Windows operating system on their web site. In this case, the Windows
advanced tab shows what the Focusrite driver set the sample rate to. The small window in the
bottom of 7-1 is where the actual sample rate is set. The Windows software will not set the
sample rate, although the bit length will be set.
Let’s look at another example where the A to D converter is located on the computers
main board. A Realtek Chip Set is used for the conversions and will respond to the Windows
software via the Control Panel settings. Refer to figure 7-2 for the results of setting the
converters’ values.
 62

Figure 7-2 Realtek Settings to 96 kHz and 16 Bits

In this example the A to D and D to A Converters Sample Rate and Bit Length has
been correctly set for the transcription by the windows “sound” section.
The next step is to have the DCart10 software connected to the converter. The
preferences tab under Edit should have the correct converter shown. Sometimes the
preference for the sound card will show a phrase primary sound driver or primary sound
capture driver. It is best to use the arrow besides the software option to select the actual
device name. Refer to figure 7-3 for connecting the Focusrite External USB converter to the
DCart10 software and figure 7-4 for the on-board Realtek chip set to be used by the DCart10
software.
63

Figure 7-3 DCart10 Preferences set to Focusrite Converter

Figure 7-4 DCart10 Preferences set to Realtek Converter

 64

Now the converters’ values are correctly set and the DCart10 software has the device
connected, the DCart10 Record Window will now be set to the same values that were
previously used for the converter.

7.6.2 Checking the Recording Window

Figure 7-5 shows the recording window (press the red button in the toolbar to bring it
up) set for 96 kHz and 16 Bits using the Focusrite Converter, while 7-6 shows the window
with the Realtek Chipset.

Figure 7-5 Recording Window with Focusrite Converter

 65

Figure 7-6 Recording Window with Realtek Chipset

The DCart10 software displays the recording variables that will be written to the header
file when it is produced. At this point, you cannot change the converters settings by only using
the recording window settings. A change via the Record File window will be written to the
created file header, but, the actual sample rate could be incorrect. Correct software control
requires that the communication between the DCart10 and the actual converter is working,
and this communication does not always work. The reference section of the Handbook details
a method used to check the converters’ actual performance if you have any doubt in the
accuracy of the settings.
In conclusion, the way to record the sound at the desired sample rate and bit resolution
is to proceed in this fixed order:

1. Set the A to D and D to A Converter to 96 kHz and 16 bits by the manufacture’s

software and/or the Microsoft Control Panel/Sound Section.
2. Set (if needed) the DCart10 sound card preferences to your converter.
3. Adjust, if needed, the values in the Recording Window to agree with 96 kHz sample
rate and depth of 16 bits.
 66

7.7 Playback System

During the transcription operation it’s important to hear the music being recorded so
that you will know if any distortion is occurring from a wrong setting and when the music
starts and stops. The level of playback should be low so that any possible audio feedback does
not occur between the recording and playback. This loopback option can be set via the
software settings in the recording properties or by using a preamplifier with multiple outputs
so that one output can go to the A to D converter and the other to the playback amplifier.
Refer to figure 7-7 for the listen while recording settings using the Focusrite converter. This
is an example of where the Manufactures software does work with the Microsoft Drivers for
loopback yet has a problem with the control of the Sample Rate.

Figure 7-7 Listen while Recording

This same method works for the Realtek Chip Set or other converters.
 67

7.8 Details of the Transcription Operation

7.8.1 Clean the Record Surface

The surface of the record should be cleaned of dust and dirt. Each type of record has a
specific cleaner that can do an excellent job and another cleaner that can destroy the record!
For example, water will clean a vinyl LP record but will damage an Edison Diamond Disc
record. The reason that water will damage the Edison records is that this record was made
using a wood powder core material that was sandwiched between plastic surfaces. The wood
core will absorb water and change the shape of the record. Alcohol will clean an Edison
Diamond Disc record’s surface but will damage a 78 RPM record due to alcohols ability to
dissolve the shellac present in the record used to bind together the rock material.

7.8.2 Set the Speed to the Value for the Record

With the possible exception of Edison Diamond Disc records (set to 78 RPM not 80
RPM), the turntable should rotate at the correct speed for the record. If the turntable cannot
rotate at the correct speed, set the speed as close as possible and refer to the later section in
this chapter on speed correction. The DCart10 help files provides a path to a printable Strobe
Disc that you can use to check the speed of your turntable by using fluorescent lights for
illumination.

7.8.3 Prepare the Software

Set the sample rate and bit resolution using section 7.6. With the Record File Window
selected and the sample rate set to 96 kHz, resolution to 16 bits, then set the recording software
to the pause state just before you are going to record the music. Refer to Figure 7-8:
 68

Figure 7-8 Recording Window Paused

7.8.4 Recording of Raw File

One method for recording the song would be to start the recording just after the needle
drops into the music groove a couple of seconds before the music starts. This ability to press
the recording button at this exact moment is hard to do and is not needed. Instead, after the
recording has been made the sound of the needle drop and noise before the start of the music
can be removed later in software. Thus, the Rec button can be pressed before the stylus has
been placed onto the record. The method used for a single song is to:

1. Press the Pause Button in the recording window.

2. Start the turntable rotating.
3. Press the Record button in the recording window.
4. Place the tonearm either at the beginning of the record or prior to the start of the music
you want to record.
5. When the music you want to record has stopped press the Stop Button in the recording
window.
6. Press the Save Button in the recording window and place the file in a temporary
location as a Raw File.
 69

For an LP or other record that has multiple songs on one side of the record you can
start and stop the recording process for each song, or the entire side can be recorded as one
big file and split apart using the software chop file into pieces option under the CD-Prep Tab.
To record the whole side containing multiple songs treat the recording as one big file using
the same steps as previously shown. The individual songs will be separated from this big file
in section 7.8.6.
While you are recording watch the Rec Level meters carefully. The meters will indicate
if the signal exceeds the maximum range of the analog to digital conversion. Since the music
has both loud and quiet passages the preamplifiers gain must be set to handle the total range
of the music and should not be changed during the recording process. The way that you can
tell if you have set the levels correctly is to always have the recorded music stay in the green
for the entire song. This is not always easy to do as noise events, which are not music, can
often cause the level to hit the red or saturate the conversion process. A good recording will
have the quiet passages at least around -10dB to -20dB down, and the loudest passages down
around -3dB. The music will have noise spikes present and these spikes are allowed to hit the
maximum level, while the music cannot. To state this an alternative way, the actual music
should range from a maximum of -3dB to below -20 dB while any noise event can hit the red
level. The -3dB upper level provides head room for the later EQ curve and file conversions.
Later figures will help to understand this relationship between the music level and noise level.
While you are recording, you must also watch the Rec Level meters to observe correct
balance between the Left and Right Channels. This process can be difficult when noise events
occur as they are often specific to a channel. The recording balance can be again checked and
corrected after the raw recording and before the original file is saved.
If the recording does overload the A to D conversion, the gain control for the
preamplifier is reduced and you repeat the recording process. Refer to figure7- 9 for a typical
display for the level meters while recording and figure 7-10 when you stop the recording after
the music has stopped.
When the Raw file has been recorded, this file will undergo several steps to create the
final Original Recording. These additional steps will be shown in section 7.9.
70

Figure 7-9 Recording

Figure 7-10 Stopped Picture

 71

7.8.5 Creating a Single Raw File

If you are ready to store this music press the Save button. At this point you will save
the file into a temporary location of your choosing for some extra work as a Raw file.

7.8.6 Creating Raw Files from a Multiple Song Recording

If you have recorded a large file that contains several songs within one continuous
recording, as would happen from a LP record, then you can split this large file into individual
raw recordings by using two DCart10 Software tools in the following sequence and shown in
figures 7-11, 7-12, and 7-13.

1. Let the large file with multiple songs become a source file.
2. Under the CD-Prep Tab use the Find and Mark Silence Passages feature on the
source file to create Markers that occur during the silence between songs. The
marker location for silence can be set manually if desired.
3. Playback the recording to verify and adjust if needed the marker locations so that
the marker is in the dead space between songs and not just during the fade-down
of the music.
4. Use the chop file into piece … feature under the CD-Prep Tab to create
individual original recording from the previous large file.
5. Change the file names for the newly created music files to correspond to the
correct names and save them as raw files.

Figure 7-11 Find and Mark Silence

 72

Figure 7-12 Markers at Silence Locations

Figure 7-13 Chop File into Pieces Command

The original use for this algorithm was to prepare a CD with individual songs from a
continuous long music file with multiple songs. That is why the Create Play List option is
shown. Splitting an originally recorded long file into separate files for restoration using this
software demonstrates how flexible DCart10 Software can be.

7.9 Steps to Create the Original File from the Raw File

Before you can save the file, you have just recorded as an Original File, some important
steps are needed.
 73

7.9.1 Checking and Correcting the Channel Balance

The channel balance, which refers to having equal amplitudes for the left and the right
channel, is difficult to perform when you are watching the music amplitude via the recording
meters. A better balance can be obtained using a software tool in the Virtual Phono Preamp
(Introduced in software version 10.80). An example will demonstrate the method. Refer to
Figure 7-14.

Figure 7-14 Left and Right Channel Out of Balance

 74

In figure 7-14, the top or Left Channel is smaller in amplitude than the Right channel
with the meter displaying the relative amount of un-balance. Since the L & R channels maybe
combined for a Monophonic Recording or Used as a Stereo recording, the balance between
the two is important to remove noise or have a correct stereo sound. The Virtual Phono
Preamp should have flat preamp and acoustical settings selected when the balance is
measured and adjusted. When the preview button is activated, the result on the meter under
the Balance setting shows the amount of imbalance present as the music is played.
By moving the Balance slider above the meter, you can reduce the value on the meter
to a neutral setting. At this point, you have a raw recording so there will be noise spikes at
times. Also, due to groove wear in the record, you may see the meter movement from neutral
at times. Set the balance slider for the best overall setting while using the preview during the
song. Refer to figure 7-15 for a balance value selected for the waveform in figure 7-14.

Figure 7-15 Balance Setting to be Used.

After using this setting, the result is shown in figure 7-16.

 75

Figure 7-16 Destination Result from Balance Setting.

After running the filter, the channels Left, and Right are very close in amplitude over
the length of the song. Because the flat and acoustical setting was selected, no frequency
modification was made to the raw recording. Proceed to Trimming and checking for overload
in section 7.9.2
 76

7.9.2 Trimming File Length and Checking for Possible Overload

Check the raw file to see if you have overloaded the analog to digital conversion
process by displaying the entire song and then looking for any music touching the very top or
bottom of the chart disregarding short noise bursts. If the music touches the top or bottom,
then the song should be recorded again, and the previously saved file deleted. Refer to figure
7-17 for an example of a good raw file. The noise spikes can touch the top or bottom as they
will be removed later.

Figure 7-17 Raw Recording

Notice in figure 7-17 how the heavy waveform is well below the upper and lower limits
of the analog to digital conversion process, while some noise bursts (spikes) are close to and
even touching the limits at times. The high levels for the noise bursts are not a problem for
the restoration process. Let’s examine in detail the raw recording from figure 7-17.
The beginning of the raw recording corresponds to the time before the stylus has settled
down into the groove (dead time) and can be removed. An example of this would be when
the stylus is moving via the turntables starting motion of the tone arm or when the stylus is
hunting to find the music groove.
This dead time maybe useful for an advanced signal measurement that is found in the
reference section where the noise of the recording system is measured. This section, however,
will be removed for all the restoration work shown.
The next interval of time is after the stylus is in the music groove and before the song
has begun. This section is important to keep as it provides an indication of the surface noise
present on the record surface and can provide the noise filters with important noise removal
information. This section should be retained for the original recording.
We can locate these various sections on the raw recording by playing back the
recording and hearing the change in background noise as we watch the playback position on
 77

the waveform. Figure 7-17 has been magnified and the beginning area is shown in Figure 7-
18:

Figure 7-18 Beginning Section

Marker 1 is in the section before the stylus has contacted the record and is quiet.
Between Marker 2 and 3 you can the sound before the stylus falls into the record groove. At
Marker 3 the noise has changed, and the stylus is in the groove. After the Marker 3 location
the display does not clearly show the actual start of the music due to noise. This section where
the music starts can be heard and identified using your ears. The section from the beginning
of the file to marker 3 can be removed so that what is left is the time when the stylus is in the
music groove and before the song has begun. The amount of time to keep for this type of
groove noise can be as short as a couple of seconds.
A similar method occurs at the end of the recorded music. When the music has ended,
the groove will spiral into the center of the record which in turn may cause the tonearm to lift
up and proceed into a shut-down sequence. As the song ends the level will often fade down
to a quiet sound with the stylus riding in the groove with no music. When the groove spirals
 78

in and if the lifting of the stylus occurs, you will often hear a thunk as the lifting mechanism
moves the stylus. Refer to Figure 7-19 for details of this ending part of the music.

Figure 7-19 Ending of the Music

Marker 1 is the location where the music has faded to a level where it cannot be heard.
Marker 2 is the beginning of the stylus lifting, and Marker 3 shows the resulting silence as
the stylus is not in a record groove. You can remove the section from Marker 1 to the end of
the file.
 79

After you have trimmed the beginning and end of the song the file can be closed and
placed into the file location for your original recordings if a flat preamplifier was used, the
correct speed was used for playback, and the channels are balanced. Place this file into the
Original Music location for the artist and record type that you have previously established
with the Music File Organization chapter.
If the speed of the turntable was not the correct speed for the record and/or the
preamplifier applied equalization, then extra steps are needed on the trimmed file before it
can be stored as an original file. Section 7.9.3 details any extra steps after section 7.9.2 have
been performed on the Trimmed Raw Recording.

7.9.3 Extra Steps to Remove EQ and/or Correct the Playback Speed

Remove Equalization from the Recording

If equalization was applied by the preamplifier during the Raw recording, then this
must be removed to produce a flat recording to become an Original File recording. You can
create an equivalent flat recording with the DCart10 Software tools.
Let’s use an example to help clarify this point starting with a raw recording of any type
of record that used a RIAA EQ preamplifier. To create a flat recording, you must undo or
remove the RIAA curve. The correct equalization for the record will be applied later during
the noise removal process; the only equalization to remove is the one that was applied during
the transcribing operation in the preamplifier.
The chapter on Equalization and Removal contains information on the method to
remove the preamplifier equalization (RIAA EQ) and that method will be repeated. After the
file has been trimmed the software filter shown in figure 7-20 will be used to process the file
with the settings shown.
 80

Figure 7-20 Removal of RIAA curve from a LP type Preamp

Notice in the options that the record type Acoustical was selected even though that may
not have been the actual record type. This software option will convert the recorded file to a
flat preamplifier type if the acoustical option was picked. We are using the software to
remove EQ with the Acoustic Setting. After the file has the RIAA curve removed from it, it
can be saved as an original file in your music file structure if the correct record speed was
used during the transcription.

Correct the Speed

The speed for recording the record should have been the correct speed. In some cases,
the speed during the transcription may have been slower than the correct speed due to either
the record speed was not available on the turntable, or a damaged record required a slower
speed to keep the stylus in the record groove. For the case where the recording was made
using a flat preamplifier DCart10 software provides a method to convert the frequencies on
the recorded file to correspond to a higher (or lower) record speed. The change speed software
(under the effects tab) for this operation is shown in figure 7-21:
 81

Figure 7-21 Speed Change

This software performs a speed change by modifying the digital values in the file to the
desired turntable speed from which they were initially recorded. Note that the change from
33 ⅓ to 78 RPM must be performed in two separate steps; first to 45 RPM then to 78 RPM
because of the operation of the software. If a lower speed was used for a 78 RPM record, the
preferred speed would be 45 RPM.
If a flat preamplifier was used, then after using the speed correction you are finished,
and the resulting file will now be saved as an Original Recording.
If any type of equalization was performed by the preamplifier and a speed change is
needed, a specific order in applying the software filters is required. This specific order is very
important to perform and to understand.

Specific Order of Speed Correction and Equalization Removal

If the speed was incorrect for recording and RIAA equalization was applied, then a
specific order of removal and speed correction will be needed.
An example will explain why the equalization must be removed prior to the speed
correction by picking a frequency within a song and see what happens to the level of this
frequency with speed change and equalization. Start with a frequency of 1000 Hz that has an
 82

amplitude of ±10,000 counts (full scale ± 32000 counts) which was originally recorded at 78
RPM on an electric recorded record. The 1000 Hz tone was in the constant-velocity part of
the frequency range and therefore no change was made to its amplitude when it was originally
recorded. Now let’s reduce the turntable speed to 45 RPM and transcribe the music with a
preamplifier that has a RIAA curve in the circuit. The original 1000 Hz tone will become
≈570 Hz due to the speed change. The RIAA EQ curve in the preamplifier will now increase
the amplitude of this tone by ≈3 dB or an amplitude of ≈ ±14125 counts. Refer to figure 7-
22 for the RIAA curve:

Figure 7-22 RIAA Curve

At this point two different outcomes are possible. If we change the transcribed file to
an equivalent one at 78 RPM by using the software speed change (figure7-21), then the 570
Hz tone is restored to the correct 1000 Hz tone but at the wrong amplitude of ±14125 counts.
When you then apply the reverse RIAA curve to make a flat recording, the amplitude will
stay at ±14125 counts since the RIAA EQ reverse curve provides no amplitude change at
1000 Hz. See figure7-23:
 83

Figure 7-23 Reverse RIAA Curve

The resulting music will not sound correct for 1000 Hz and many other frequencies.
Let’s perform the techniques in the correct order. While the music file is still at the lower
initial speed of 45 RPM, we will apply the reverse RIAA EQ curve. This will take the
amplitude of the 570 Hz tone down to ±10,000 counts since the curve is a mirror image to
the RIAA curve in the preamplifier. Now when the speed change is made to 78 RPM, the 570
Hz tone becomes 1000 Hz at the correct amplitude. Therefore, the order of applying the
equalization is very important.
The specific sequence is to first remove the equalization before any speed correction
has been made. After the equalization is removed from the recording (at the speed it was
transcribed) the file is now equal to a flat recorded file. Next, the speed correction can be
made to bring the music tones to where they would be if they had been played at the correct
speed. This method is the only way that can be used for a successful conversion of this type
of raw file to an original type for music restoration.

7.10 Final Preservation as Original File

The original file should always be preserved as it is. Make a copy for any working file.
 84

8.0 General Noise Removal Methods & Level Creation

This chapter covers the general methods used for removing noise from the music file.
The result of your noise removal work will be to create both a Level One and Level Two music
files. Later chapters will describe specific noise removal methods for different types of
records.

8.1 Definition of Noise

For music restoration noise will be anything that has been added to the original
recording that was not wanted. The distortion that occurs to the music from record wear will
be included in the noise term.
In general, we refer to aspects of noise using many different terms to qualify the type
of noise. For example, in the DCart10 software there are terms for noise with a specific word
used to describe the character of the noise. Software terms used to describe this noise as
impulse, crackle, narrow crackle, and big click. A further distinction is made between impulse
noise and continuous noise. Impulse noise is any type of interruption in the music that is of
short duration. An example of this would be the sound that a crack in the record makes when
played. Information on these descriptions can be found in the help files with the DCart10
software. The other general type of noise is a continuous type that is always present though
out the music. A good example of this type of noise occurs just before the music starts and
continues throughout the song. These two general types of noise will be dealt with by two
different Multifilters, namely Part One and Part Two.
Another important aspect to record noise is that the shape of noise is unique for each
type of recording media. The shape of the noise waveform that you find on a 78 RPM
electrical record has a different shape than that from an Edison Diamond Disc record. The
stylus motion for the vertical cut records gives a different noise signature than the lateral cut
motion. While the shape of these noise events is unique, they do share some similarities. The
noise software filters while have unique settings for different types of recordings.
Multifilters are a useful concept within the Diamond Cut Software options that allow a
series of individual filters to be strung together into an overall block. While it is possible to
perform each operation in a separate step the multifilter concept has several important
advantages. One advantage is that all the operations in the block are performed as digital
words in a continuous mathematical operation whereas performing each operation separately
involving mathematical conversions. Another advantage is that you can review and see your
 85

total operation on the music in one saved filter containing all the unique settings for each
individual part.

8.2 View used in DCart10 Software Display

DCart10 software has two modes for viewing and editing the audio data; namely classic
edit and fast edit. The classic edit mode will be used for all editing work in the handbook as
the ability to view both the source and the destination at the same time while removing noise
will be a Key to the success of the noise removal process. This advantage will be shown later
in this chapter under the Sight & Sound section.

8.3 Details of Software Noise Removal Algorithms and Sample Rate

It’s important to state that the specific algorithms that the DCart10 Software uses to
remove the noise are trade secrets and are not known. The information that has been learned
from using these filters will be described but the specific inner workings of the software are
not known.
After the restoration of many songs, the sample rate of 96 kHz for the music was
determined to allow the noise filters to work at their peak performance. If you have a music
file that was not at this sample rate, then the DCart10 software will not remove as much noise
as a higher sampling rate could do. Also, the noise filters can usually be set to a more
aggressive setting at 96 kHz than a lower sampling rate.
If your original file is at 44.1 kHz you can still achieve excellent results, but the
restoration work may take longer as you may have to remove more noise manually.
The DCart10 software provides a method to increase or decrease the sample rate and
bit depth via software algorithms however, this is not a substitute for an original recording at
the higher rate. If the source file was originally recorded at 44.1 kHz, then the noise should
be removed at this rate and not converted to a higher value. Useful noise information cannot
be added to the original recording via an algorithm that increases the sample rate or bit depth
as the algorithm has no knowledge of the noise.
 86

8.4 How Much Noise can be removed?

This question is important to understand. Any software algorithm that removes noise
from a music file does not have the ability to consistently remove just the noise and not some
of the music. All the different software algorithms must make difficult decisions on millions
of digital bits and while they do use very clever engineering techniques, the algorithms will
fail at times. The use of these noise filters can remove some of the desired music so that you
must be careful on how to apply them. That is why when the noise filters are being used, it is
very important to listen to the music before and after the filtering to determine if your settings
were too aggressive and caused some of the music to be removed. When I first started to
restore music, I would set the filters to a very aggressive setting so that after just one pass
through the music I could not hear any noise. Later, I realized that I had also removed some
music!
A universal setting for the noise filters is not possible due to the variation in the
condition of the records and how the content of the music affects the performance of the noise
filters. The best method to remove the noise is to start with the filter settings set to a low level
and then try a pass through the music. If you have a relatively small number of noise events
left after using the filters, then the remaining noise can be removed by a manual technique
called the Sight & Sound method. While very aggressive filter settings can give the
impression that the noise has been removed; some music will have been removed with the
noise. It may not seem correct that you should be removing noise manually when you have
software that can do the removal for you; however, a balance is needed between time to
manually remove noise and removal of desired music.
There will be records that have so much noise that the ability to remove it is just not
possible. There were many millions of records produced and if the recording you have is not
cleaning up as well as you want then it is best to stop at what you can achieve. You can often
find better versions of the recordings as you continue to add to your collection.
Brass instruments can create a music shape that looks like a noise event instead of
music. There is a box that is checked in all the filters to help avoid the removal of brass sounds
from the music. When filters start to remove brass notes, the sound becomes somewhat mussy
instead of having a sharp and clean sound.

8.5 General Noise Pictures

For a starting point to the noise removal process let’s examine a copy of an original
song file before the use of any noise filters. This working copy is placed as a source file. The
 87

transcription was performed with a sample rate of 96 kHz and a digital word length of 16 bits.
The file is displayed using DCart10 in classic mode. Refer to Figure 8-1 below.

Figure 8-1 Original Song File

When the total song is displayed the detail is hard to see, however, there are several
large spikes that show a sudden changes in the amplitude of the music. Specific sections will
be zoomed in using the software tools to see more details.

8.5.1 Magnified Picture of Noise

Figures 8-2 and 8-3 magnify some of the noise events for more detail.

Figure 8-2 General Noise in Music

 88

Figure 8-3 Magnified Detail on Noise

Notice the shape of the noise with the fast amplitude changes and the differences in the
shape between the right and left channel at the same time.

8.5.2 Use for the Section before the Music Starts

Figure 8-4 is a zoomed in look that shows the moment when the stylus is in the record
groove and the music starts. The location between Marker 1 and Marker 2 in figure 8-4 shows
this time before the music starts. Note the recording shows activity before the music starts
which is the surface noise of the record. The reason that the waveform line doesn’t appear as
thick as it does in this same area in Figure 8-1 is that the time scale is different between the
zoomed part vs the original and now the signals are not bunched up in time.
 89

Figure 8-4 Before the Start of Music

There is noise before the music starts which comes from the fact that the surface of the
record has a texture to it that causes the stylus to move in a somewhat random motion. This
section just before the music starts is used by the EzImpulse Filter in its noise removal
algorithm. It is important to keep the record noise before the music starts for a short amount
of time (couple seconds or more).
This section before the music starts has another use during the application of the
Continuous Noise Filter (CNF) in the Level Two File Generation. The methods shown later
for specific CNF settings start with a general setting that is effective for most records. A
change to the CNF from the general settings can occur in the Level Two file creation using
this period from Marker One to Two as a new sample for the CNF for a specific song.
Later in the final part of the restoration process, we can remove this section of the
recording before the music starts.
 90

8.6 Noise Masking Technique Beginning and End

There is a useful noise reduction method that can be used when the music is loud,
namely the ability of the music to over-ride or mask the noise. Many songs have a constant
level of background noise that is always present on the song. Records made prior to the use
of vinyl material have a built-in noise that is always present and quite noticeable. The way
that the ear hears and processes sound though, can be used to reduce this background noise;
namely that when the music is loud, your hearing cannot distinguish the background noise
from the music. Therefore, when the stylus first settles into the groove and the music has not
started, the listener will hear the background noise and be somewhat conditioned to having
this noise present in the recording. When the beginning of the song starts, often the music is
loud and now the background noise cannot be heard…however… the listener had heard the
noise earlier and to some extent that impression will stay regarding the quality of the song.
As explained in section 8.5.2, this noise before the music starts will be useful for noise
removal, however when the final restored version is ready this part will be removed. By
removing this starting noise, the overall quality of the music will be increased. If a moment
of silence is desired prior to the start of the song, then the software can add silence that is void
of any noise.
Another important aspect to record noise and the ability of the music to mask its
presence occurs at the end of many songs during the common fade down ending. Many songs
do not simply suddenly stop. Rather the volume is slowly reduced to the point where you
hear both the music and background noise at the same time. Now the record noise that is
always present will stand out. One method that helps reduce noise at this point is to add
additional fading to the music over and above the original fade down performed by the
recording engineer. Since additional fading at this point reduces both the music and the noise,
the overall effect on the noise presence can be improved. You must use careful judgement
applying this additional fading as you do not want to remove a part of the song when the
music is still prominent.
An example will be used to demonstrate this method. Figure 8-5 shows a song with a
strong presence of background noise remaining while the music is still playing during the
slow ending.
 91

Figure 8-5 Ending of the Song

At the point where the vertical black line is located, the level of the music has decreased
to a point where the noise dominates the music. The music is still undergoing a reduction
(fade down) so that a simple stopping of the song at this point (vertical black line) would not
sound natural. To improve this song the fading after the black line will be increased so that
the highlighted area (lighter area after the line) undergoes more reduction. The Diamond Cut
Software edit command fade down is used. Refer to the figure 8-6:

Figure 8-6 Fade Down

 92

This software option shows a line indicating the reduction in the signal level and thus
a fade down. The result of the use of this signal reduction is shown in figure 8-7:

Figure 8-7 Fade Down Result

When the song is played the additional reduction in level at the end of the song will
still seem natural and the background noise is further reduced. The value of 20 dB was found
to provide a smooth ending to the music.
The use of both noise removal prior to the start of the music and extra fading towards
the end of the song will be performed at the end of restoration steps during the Level Two file
creation. This will be explained in the Level file discussion.

8.7 General Noise Steps for all Types of Recordings

The removal of noise for all records involves several similar steps. These Multifilters
have names Part One and Part Two to aid the proper sequence of noise removal. The use of
the terms Level One and Level Two describe two sequences of using the Multifilters along
with additional steps. The Levels are stored as separate results of noise removal in the
Restoration Level Folder. For 78 RPM Electric recorded records the Multifilters used for
noise removal will be 78 Electric Part One and 78 Electric Part Two. For other types of
records, a similar labeling process will be used. Detailed steps will make this concept clear
for the noise removal method.
 93

The reason that the noise removal process will be performed in two levels of operation
is due to the nature of the noise, the variation in record equalization, and the operation of the
software along with the ability to return to previous work.
As seen in figures 8-1 to 8-4 there are two general types of noise: impulse of short
duration and continuous noise. The DCart10 Software has specific tools to deal with these
types of noise and they are different in operation and will be performed in separate parts of
the operation which are called Part One Multifilter and Part Two Multifilter. Part One of the
filter performs operations that do not remove significant frequency components of the music.
Part Two can modify the frequency content of the music and uses the CNF (continuous noise
filter) The CNF works best after the impulse noise has been removed in Part One and will be
performed during Part Two.
Another reason to separate the use of the Multifilters into two specific parts is due to
the ability to see and hear the location of the noise event. The use of the CNF causes the
background noise to be reduced and the introduction of a noticeable time lag between events
in the source file and the destination file. This time lag increases the difficulty in finding the
exact location for the noise during the Sight & Sound operation that is later described. Also,
the reduction in the background noise causes small impulse noises to be harder to detect with
your hearing.
The desired record equalization will be applied after a Level One File has been created.
The creation of the music in the Level One folder requires the most time for your restoration
due to the potential need for manual noise removal. Since the application of EQ to the record
cannot be removed in software for all but RIAA and NAB Tape curves, we can always return
to the Level One file and try another EQ without redoing the labor-intensive manual noise
removal operation.
An important sequence of operations must occur, which is somewhat confusing at
first. Keep in mind that the Levels refer to a major operation while the Part One and Part
Two refer to specific Multifilters that are used within the Levels.
This order of noise removal operations will be repeated for each type of record within
the Level One and Level Two file generation sections shown in later chapters. Let’s start by
listing the general order of operations for all record types:
 94

1. Copy the Original File (Flat Recording) and use it as the Working Source File. Check
the channel balance and if needed change gain of a channel using the balance meter as
shown in section 7.9.1.
2. Apply the Specific Multifilter Part One to the source to create a destination file starting
with the values shown for the specific records. If many noise events remain in the
destination file, increase the settings until the noise has been removed or the music
starts to be degraded. Change the EQ settings (if used) for the desired sound.
3. If needed, manually remove noise that the Part One Multifilter could not remove from
the source file. This operation is called the Sight and Sound Method.
4. The Source file in step 3 is now saved and this file is a new Level One file. The
previously created destination file is not saved.
5. If noise removal for the working source file was not satisfactory, you can save the
source for another time as a Level One A file.
6. Copy the Level One file and use this file as a working Source file.
7. Apply the Part One Specific Multifilter again to the Source.
8. Convert the resulting Destination file to a new Source file.
9. Change the 96 kHz sample rate to 44.1 kHz sample rate (or keep at 44.1 kHz if this
was used for Level One).
10. Normalize the new file to have a maximum value of 0.0 dB.
11. Apply the Part Two Specific Multifilter to the new normalized Source from step 10. If
needed, change the CNF settings for more noise removal while preserving the music.
12. Trim away the non-musical section of the music files beginning and end.
13. If needed, add additional fade down to the ending of the music.
14. Save the destination result of step 13 as a Level Two music file.

The DCart10 software uses high frequency information to remove the noise. A flat
recording will have the greatest amount of high frequency noise since the high frequencies
present on the recording have not been modified from an EQ (equalization) curve. For both
vertical and horizontal records, the first stage in the Part One Multifilter is a De-Clipper to
aid in removing large noise events. In the Part One Multifilter for lateral records the second
operation is removal of the noise, conversion to a Monaural file (no conversion if a Stereo
File), and then application of EQ to make the music sound natural as we play the destination
file. In the Part One Multifilter for vertical records the second operation is conversion to a
Monaural file followed by noise removal.
If the later Sight and Sound Editing is needed, it is easier to hear, find, and remove left
over noise with manual methods when the music sounds natural. A thin or shrill music
 95

passage can cause your hearing to have a hard time detecting the crackle or click that a noise
passage makes.
Various EQ options maybe used while listening to the results for the song and the
desired one will be used to create the Level Two file. Remember that while the EQ is used in
the Part One Multifilter, the resulting Level One file will still be a flat file since the Source
will be saved as the Level One file and not the Destination (Step 4). The Part One Multifilter
and Part Two Multifilter are both used to create a Level Two file.

*** Important***

The fact that all the Level One files came from a Flat Recording means that possible
changes to the EQ can be later made without losing your noise removal work by repeating
the generation of a Level Two file.

8.7.1 Creation of a Level One File

The result of these steps will create a music file for the folder Level One so that when
the noise removal multifilter Part One is used on this file, a clean (as possible) file will remain
that needs no additional impulse noise removal.
Another way of stating this is that if you have spent time manually removing noise
events (that could not be removed with the automatic software filters), you want to be able to
return to this same file in the future as your knowledge increases about noise removal without
losing the time you spent manually removing some noise events. Therefore, if you keep the
music file containing noise that can be removed at this time with your noise filter and in the
future improve the filtering even more, you can return to the same starting point with your
previous manual noise removal work saved.
An example will help to clarify this idea. In the following song the source file contains
a completed Level One file. The destination contains the result of applying a noise removal
Multifilter Part One. The section shown in the source contains two channels and a noise spike
just before Marker 1. The Destination contains a single channel and the removal of the spike.
Refer to figure 8-8:
 96

Figure 8-8 Noise Removed from Source

All your files that are in the Level One folder will have impulse noise removed only
when the corresponding Part One Multifilter is used on them. If the noise could not be
removed at first by the Part One Multifilter then some extra manual removal is needed to
create the Level One file. A final Level One File requires ONLY the application of the Part
One Multifilter to produce a noise free result.
 97

Manual Noise Removal Sight & Sound Method

After the DCart10 noise filters have removed as much of the noise as possible without
removing the desired music and some noise remains, a manual method is used to finish the
removal of impulse noise for the Level One file. The manual method uses a combination of
sight & sound. The original file is placed in the source file location using the classic edit
option for file view. Next a specific multifilter for the type of record is run and the result is
shown in the destination window. The multifilter used will be the Part One Multifilter. The
next step will be to find any noise events that are left in the source from the use of this
multifilter and remove those noise events using manual methods. The result of automatic and
manual noise removal technique allows the desired Level One file to be produced.
The manual removal of the noise events works well since when your ears hear a noise
your eyes will then see the specific location of this noise. This method requires that the source
and destination files be in sync so that the specific location at the source can be removed
when you hear the noise in the destination. In the DCart10 software there is an icon in the
toolbar to Sync the source and destination files, which should be selected.
Note: Even when the Sync icon is selected, the use of some filters will cause a time lag
between source events and destination results. The filters selected in the Part One Multifilters
do not create a noticeable time lag.

The Sight & Sound manual noise removal process consists of the following steps:

1. Keep the Multifilter window on the screen and as far down as possible but still able to
press the Run Filter button.
2. Starting at the beginning of the destination file, carefully listen to the music while
watching the cursor move. When a noise event is heard, immediately stop the playback
of the music (press spacebar or stop button) and locate the source of the noise in the
destination window. When the noise in the destination window is found, look above to
the same location in the source window using the playback cursor. You will be able to
locate the specific noise event that the multifilter did not remove in the source file. The
zoom in button and highlight section along with the replay of the zoomed in destination
section can be performed several times to work down to the specific source of the noise.
Remove the noise event in the source by using the manual interpolation “I” key or other
manual methods.
3. After the noise event in the source has been fixed continue to play the music in the
destination window just after the last noise event or re-do the multifilter again on the
 98

edited source (remove the previous destination file first and press the run filter button)
and continue to listen for more noise and if needed remove additional noise events from
the source file.
4. When the destination file has been completely heard and the remaining noise events
removed to your satisfaction the edited source file becomes a Level One File Creation.

Let’s look at an example of this sight & sound method using the 78 Electric Multifilter
Part one with a 78 RPM Electric Recording. Figure 8-9 shows the result of the first pass on
the 78 Electric recording from the original file. The source is the original recording at 96 kHz
and 16 bits word length, and the destination contains the result of the filtering.

Figure 8-9 Multifilter Part One Run, Ready for Sight and Sound

Some detailed remaining noise pictures are in figure 8-10.

 99

Figure 8-10 Noise Events in the Destination\Source Windows

The marker shows the noise event in the destination window that was not removed by
the 78 Electric Multifilter Part One. The corresponding source file location shows the noise
event that you want to manually remove (since the multifilter could not). Figure 8-11 shows
the noise event in the source zoomed in and highlighted prior to the use of the manual I key.
 100

Figure 8-11 Noise Events Prior to Removal

The highlighted section is where the software will remove the offending noise event
by using the I key. The result of using the I key to remove the noise is shown in figure 8-12:
 101

Figure 8-12 Manual Noise Removal Results

Notice that the noise event has been removed from the source, but the noise is still
shown in the destination file. The noise removal from the source can be verified by removing
the previous destination file and re-running the Part One multifilter and hearing the music in
the new destination file. An alternative method is to advance the play back position on the
destination file slightly after the last found noise event and continue to play the music and
listen for another noise event. By jumping ahead of the last noise, you can save some time by
not waiting for the multifilter to run again after the last manual noise removal.
Some caution is needed with the manual method of removing noise since the use of the
I key requires that you place the interpretation range (highlighted in Figure 8-11 and 8-12)
 102

around the offending noise with the correct interpolating interval. The best method for manual
removal is to always check the results of the correction by looking before and after the use of
the key to verify a smooth replacement of the music’s waveform. You can use the undo key
and try a slightly different area for the interpolation range if the result isn’t smooth and
natural. While the manual method of noise removal may at first seem difficult to use it will
become second nature in the future as you use it. The I key works over a large range of noise
types.
This process continues until the source file has had all the noise events removed that
the multifilter could not remove. At this point, the source file should contain impulse noise
that can be removed solely with the 78 Electric Multifilter Part One (for our example) except
for continuous noise. A good check would be to run the Part One step one last time to check
for any noise that you may have missed in the last pass.
Save the Source in the Level One location in your music folder. The file will remain at
a sample rate of 96 kHz and 16 bits of word length. The source now had extra noise removed
by the Sight and Sound method.

Help in Finding the Noise during Sight and Sound

The source files that you are using during the manual noise removal operation have a
lot of variation in amplitude since they are original files from the transcribing operation. At
this early stage in the music restoration, you will want to keep the source file intact and not
change its amplitude. Since you are creating a temporary destination file that will not be saved
for the future; you can change the amplitude of the destination file to help both your ears and
eyes find where the noise impulse is located for removal. While there are two methods that
you can use to increase the music volume to aid in hearing the noise, the second method will
also help in seeing the noise. The first method is to increase the music level by turning up the
volume of the playback amplifier. This is easy to do and often helps in finding the noise event.
The second method helps in seeing and hearing the noise by applying a gain change to the
destination file. The gain can be changed under the CD-Prep option in the toolbars after the
destination file is selected. The increase in amplitude of the destination file can range up to
where the peaks just touch the maximum values (0 dB).
After you have removed all the noise from the source file through the Part One
Multifilter and if needed manual methods, you now have your Level One Music File. Save
the source file in the Level One folder and continue to the Level Two creation.
 103

Why the Manual Noise Removal Method Occurs at a Specific Step

The manual removal of noise could be performed at any time on a music file. Indeed,
you may hear an annoying noise event in a music file that was in the Level Two folder or
even the Restored folder and wanted to remove the noise at that part of the restoration effort.
While it is tempting to remove this newly heard noise in a later part of the restoration work
sequence, you should only remove impulse noise using the manual method during the creation
of the Level One file .This sequence of noise removal during the creation of the Level One
File is best for important reasons that are related to the potential use of the batch editing
operation and the response of filters.
The sequence that we use for noise removal is structured so that you can return to the
Level One file and then re-run the filters to create the Level Two file. The operations during
the creation of a Level Two file can often be automated in a batch process (under the filter
tab) if any noise that needs manual work has been previously done. If you remove the noise
later than the Level One file you will lose this ability to just re-run the filters. The next reason
is related to the nature of the response of the filters.
The impulse noise from the record surface can cause a rapid change to the music
waveform’s shape. Music frequencies range from 20-20,000 Hz and have a relatively smooth
shape to them. This is in sharp contrast to the fast shape changes seen with impulse noise. In
electrical and mechanical circuits, a fast change in the shape of the signal that is applied to a
circuit will cause the time domain output response of these circuits to have a shape that is
related to the type of elements within the circuit. If the circuit contains components that can
oscillate then the time domain response can show a ringing related to these circuit values.
This is analogous to a sharp push on a swing that will then continue to move back and forth
for a while.
An example in music for us would be when a crack occurs in the record and the
resulting playback noise impulse shows a ringing as the stylus oscillates for a short time. This
ringing can cause the waveform in the source file to have a longer distortion length to it than
the original impulse itself. This initial ringing will continue into the multifilter with even more
ringing with resulting music waveform distortion. For a picture of these effects, refer to figure
8-13:
 104

Figure 8-13 Large Noise

The original impulse on the record was modified by an electrical-mechanical circuit

(stylus and cartridge) during the conversion from a crack on the record to the music source
file. The stylus in the cartridge produced some ringing that starts at marker one location on
the upper source file display. This mechanical ringing was then converted to an electrical
signal and displayed as a ringing waveform.
The equalization filter in the Part One Multifilter will also show a ringing at a
frequency related to the elements in the EQ filter. Notice in figure 8-13 how the destination
result from use of the EQ in the multifilter (lower waveform) shows the waveform continuing
to wiggle until marker two (both have the same time scale). The impulse noise that started at
marker one in the source file continued till marker two in the destination file. This extra
stretching of the noise event in turn adds distortion to the music for a longer time than just
the initial noise event. When the impulse noise is removed in the source file (location one)
then the extra distortion shown in the destination file (over to location two) will also be
 105

removed when the multifilter is again applied. The removal of noise in the source file is much
more effective than trying to removal the original event and the resulting additional distortion
in the destination file.
When you are listening for noise in a section of the music, the ability to hear the noise
and see where it is occurring is easiest before the shape of the noise has been changed.
Therefore, the noise is removed from the source file during the application of the Part One
Multifilter and not after additional filtering with the Part Two filter. If you waited to remove
the noise after several filtering operations have occurred, then your ability to find and remove
just the noise is more difficult. After the application of equalization filters and other filtering
(CNF) the noise will blend into the desired music waveform and be harder to find and remove.
The use of EQ (Equalization Filter) in the Part One Multifilter during Manual Noise removal
is a compromise to help your hearing locate noise by making the music sound natural.
If you hear some noise in your music after you have created the Level One file, the best
thing to do is to go back and create a new Level One file while you remove the newly found
noise.

8.7.2 Creation of a Level Two File

Both Part One and Part Two Multifilters will be applied to create the Level Two File.
The previous file in the Level One location will have no impulse noise remaining when the
Part One Multifilter is applied to the file. The Part One Multifilter may perform frequency
modification to the music due to a possible EQ stage. Before the Part Two Multifilter is
applied, listen to the sound of the destination file after the Part One Multifilter is used and
change the EQ if needed for the best sound.
The Part Two Multifilter will contain several CNF (continuous noise filter) devices
having a general type of response that can be applied to a music after processing by the Part
One section. If continuous noise remains with the general settings, then different settings can
be used after taking a noise sample from the beginning of the file.
The sequence of the Level Two operations will be removal of the impulse noise using
the Part One Multifilter, conversion if needed to 44.1 kHz samples per second, Normalization
of the amplitude, removal of noise that is continuous or present throughout the entire music
using the Part Two Multifilter, and then the final editing steps. During the final editing steps
the noise before the music starts will be removed and if needed extra fade down near the end
can be performed. The results of this section can be the final stage of the restoration process
 106

if you are not going to enhance the music. The subject of enhancing the music is covered in
a later section.
Because you have the previous Level One file in your Restoration Files Folder you can
always return to the creation of a new Level Two file later for any reason. Let’s apply the
steps with an example.

Apply Part One Multifilter

The effort starts by applying the Part One Multifilter to your Level One music file.
Refer to Figure 8-14 for an example using the 78 Electric Part One Multifilter:

Figure 8-14 Example Multifilter Part One Used

The destination file contains the music with the impulse noise removed and the
continuous noise remaining. As you listen to the destination file verify that the proper EQ
was used for the music. If you need to change the EQ for this song then make a temporary
 107

modification to the EQ in the Part One Multifilter, delete the previous destination file, and
run the filter again to create the desired result. The destination file becomes the source file for
the next effort.

Rate Change & Normalizing Amplitude

The previous Part One Multifilter section used a high sample rate of 96 kHz for the
music file. This high rate is not needed for the rest of the restoration work and a 44.1 kHz
sample rate works very well with the CNF to create narrow bin sizes for a given FFT size.
The 44.1 kHz sample rate when used with a digital to analog converter can accurately produce
audio frequencies from 20-20,000 Hz which spans the range of human hearing. Therefore,
the sample rate will be changed down to 44.1 kHz using the Change Sample Rate Option.
Refer to figure 8-15:

Figure 8-15 Change Sample Rate

After the sample rate has been changed the file will be the new source if the option to
open file after converting is checked (see figure 8-15).
Next, the files amplitude is normalized so that the signal levels are increased to a
maximum value of 0.0 dB. There are two methods to normalize under the CD-Prep location.
Either one can be used with a setting of 0.0 dB. Refer to figure 8-16 to see the result of the
normalization.
 108

Figure 8-16 Results of Normalization

When the file has been normalized the CNF settings will be useful over several music
files without a need to make adjustment for amplitude differences since all the files will have
had the same maximum (0dB) value. The CNF options will be selected with a normalized
music file in the Part Two Multifilter. For most restorations, the CNF settings with the specific
examples shown later will be adequate.

Apply the Part Two Multifilter & Trimming

After the file has been normalized the noise that is continuous will be easily seen. Refer
to figure 8-16 to see the noise prior to the start of music (marker 1 is the start of music). If
this noise was not present, then the waveform would show a straight line with zero amplitude
prior to the start of music. The Multifilter Part Two is now selected and run on the new source
file (Normalized and Part one applied). Refer to figure 8-17 for the results:
If the music still has noticeable continuous noise present after the Part Two Multifilter
is used, you can modify the various CNF settings and if needed sample the noise before the
music starts to assist in setting the values. Keep in mind that the use of the CNF algorithm
needs careful adjustment so that music is not removed along with the noise. The CNF settings
shown later in the record chapters are quite mild.
 109

Figure 8-17 Using 78 Electric Part Two Example

The destination file in this example shows a reduction in the continuous noise present
before the music starts.
After the 78 Electric Multifilter Part Two has been applied to the source file the
destination music file needs one last process: namely the final trimming operation. During
this part, the remaining noise prior to the start of the music and after the music is removed
(Using Control & X on a highlighted section works well for removal). The resulting file is
now placed into the Level Two file location. Refer to figure 8-18 for the final file. If wanted,
the fade down from section 8.6 can also be performed to create the final Level Two file.

Figure 8-18 Level Two File

 110

If desired a brief period of silence can be added to the beginning and/or the end of the
music using DCart10. This period of silence can be useful when selections of songs are placed
on a CD or another device to have a clean break between music.
A short one second period of silence before and after the music will be useful for
creating both playlists and making CDs of your music. Some of the CD burners will also
provide a method to add the silence. The DC Tunes software will benefit by having a distinct
silence section before and after the music as it moves automatically from song to song.
Figure 8-19 shows the DCart10 software option to add silence at either beginning or
end of the music. This software is activated only when a file is present in the source location
and is found under the Paste option from the main Edit location.

Figure 8-19 Insert Silence Software

8.9 Specific Multifilters for Records

Later Chapters will describe Multifilters for different types of records. Each will have
some variations in their makeup for the different types of recording methods. For example,
the creation of a monoaural file before the use of the noise filters works well vertical
recordings, while the conversion to monoaural is best after the noise filters for lateral
recordings. This change in location of the conversion was found through experimental tests
with many recordings.
 111

The Part Two Multifilter will use one or more CNF devices. These different types were
optimized for the type of surface noise on the type of records, but you are encouraged to
modify the settings.

8.10 Creating a Final Restoration File

After you have finished with the restoration of the file and it’s in the Level Two folder,
you often will be finished with your work. The same music file in the Level Two folder can
be copied into the Restored File location and enjoyed as is. Later, you may want to add some
enhancements to the music or try other effects that are available in the software. These extra
steps will use the Level Two file as a base for your work and by keeping the Level Two file
in its original cleaned up and equalized form you can always return to it if the enhancement
work does not work.

8.11 Method of Keeping Track of Your Restoration Work

As you are creating various Levels you will want to keep track of what files have been
worked on and what files need to be worked on. There is a feature in the DCart10 software
that can be used to help keep track of your progress. The method uses the presence of a file
that the software creates to display the music, the PKF file, to track your progress.
When the DCart10 software opens a music file the software creates another file with
the same name but a different extension. The new extension has the letters pkf. In figure 8-
20 the open software dialog window shows the files before the new file is created:
 112

Figure 8-20 Dialog Box before PKF is generated

When a file is opened by the software and displayed on the screen a new file is
generated that is used to create the waveform display. The software will store this new file at
the same location as the previously opened music file. Refer to figure 8-21 to see this new
file with the pkf extension for the song Rovin’ Gambler after the file was opened:
 113

Figure 8-21 Dialog Box after PKF has been generated

The DCart10 software will create a corresponding pkf file when a destination file is
also saved. The software will not save a pkf file for the destination when it is displayed on
your computer after action on the source file unless you save the destination file.
When you have completed the noise removal on the original file and it is ready to be a
Level One file the presence of the new pkf file can remind you that this file work is done. The
lack of a pkf file can be used to show you that the file needs noise removed since the pkf file
is only created when DCart10 opens a file.
You can modify this method to keep track of which music files have been processed in
later stages of you work. The pkf file can be deleted at any time because the DCart10 software
will generate a new file if needed for a waveform display when you open any file.
The default setting for the open file dialog box is to just show wav type files so you
should select all the “All files setting” (bottom of figure 8-20 and 8-21) to see if you have any
pkf files present in a folder.
 114

9.0 78 Acoustic Noise Removal Methods

This chapter covers the details for removing noise from 78 RPM records that used
music power solely to record the record. The term 78 Acoustic will refer to these records. The
creation of the Level One and Level Two music files will follow the same workflow as
described in the chapter General Noise Removal and will not be repeated in detail here. For
the 78 Acoustic records the same multifilter for both 44.1 kHz and 96 kHz sampling can be
used. An overall workflow is:

9.1 Level One Creation

1. Select classic view for the Diamond Cut Software.

2. Copy the desired music file from the original file folder and open the copy as a source
file.
3. Check Channel Balance and Adjust if needed using Virtual Phono Preamp.
4. Run the 78 Acoustic Multifilter Part One.
5. If needed, increase the amplitude of the destination file by using the CD-Prep options
on the destination file.
6. If many noise events remain slowly increase the EzImpulse settings to remove more
noise while checking that music has not been removed.
7. Perform the Sight & Sound manual noise removal method for any remaining noise
impulses in the source file.
8. When finished, store the source file as a Level One music file.

9.2 Level Two Creation

1. Copy the new Level One music file and open the copy as a source file.
2. Run the 78 Acoustic Multifilter Part One and make the destination into a source file.
3. Apply a Change Sample Rate operation if needed from the initial 96 kHz to 44.1kHz
to the source file. Make the result a new source file.
4. Normalize the source file to have a maximum amplitude value of 0dB.
5. Apply the 78 Acoustic Multifilter Part Two to the source file, and make the resulting
destination file the new source. If continuous noise remains, carefully increase the CNF
settings while checking that desired music is not removed.
 115

6. Trim away the remaining noise before and after the music from the source file. If
desired, add extra fade down near the end of the music.
7. When finished, store the resulting file as a new Level Two music file.
8. After the Level Two music file has been created, you can choose to copy the file into
the restoration music file location, add silence to the beginning or end, or continue to
the Advance Section.

9.3 Details of the 78 Acoustic Multifilter Part One

The 78 Acoustic Multifilter Part One is shown in figure 9-1:

Figure 9-1 78 Acoustic Part One Multifilter

The filter is composed of six parts that are detailed below with their settings. The first
filter helps to shape the noise spikes so that the EzImpulse filters can recover the music in the
noise. A progression of Impulse Filters helps to find and remove noise.
116

Figure 9-2 DeClipper (1) 78 Acoustic Multifilter Part One

Figure 9-3 EzImpulse Filter (2) 78 Acoustic Multifilter Part One

117

Figure 9-4 EzImpulse Filter (3) 78 Acoustic Multifilter Part One

Figure 9-5 EzImpulse Filter (4) 78 Acoustic Multifilter Part One

Note that the various settings are on the low side and the Solo/Brass option is on.
 118

Figure 9-6 File Conversion Filter (5) 78 Acoustic Multifilter Part One

The two separate channels are converted to a monaural file at this point. When the two
stereo files are converted to a monaural file the original stereo 45-degree motion that contains
horizontal and vertical information can be processed to remove either type. Thus, when the
two channels are added together the vertical component is removed and the horizontal
remains. The reason that the file conversion is performed after the initial noise is removed is
that the shape of the noise is greatly modified after the two channels are combined and as
such, the filters work better when they see the original shape of the noise as it directly comes
from the record, rather than after it has been converted to a monaural signal.

Figure 9-7 Bandpass Filter (6) 78 Acoustic Multifilter Part One

 119

The last filter limits the overall bandwidth of the music. The limits are very broad at
this point since additional limiting will be performed later during Part two. Some of the very
early records have frequencies higher than often stated and a lower value may needlessly
remove music.

9.4 Details of the 78 Acoustic Multifilter Part Two

The 78 Acoustic Multifilter Part Two is shown in figure 9-7:

Figure 9-7 78 Acoustic Multifilter Part Two

The multifilter part two contains three CNF sections. The use of three CNF sections
helps to remove some of the high frequency noise that the acoustic records have versus the
electric recordings. The recording engineers did not remove some of the high and low
frequency noise since the playback equipment at the time had a very limited response. With
the current playback systems used today extra music frequencies can be heard and using three
CNF sections helps to bring out the hidden music. The details of each CNF are shown in
figures 9-8, 9-9, and 9-10 below:
120

Figure 9-8 CNF (1) 78 Acoustic Multifilter Part Two

121

Figure 9-9 CNF (2) 78 Acoustic Multifilter Part Two

 122

Figure 9-10 CNF (3) 78 Acoustic Multifilter Part Two

This completes the noise removal process for many 78 Acoustic Records that were in
average condition. The CNF filters shown in figures 9-8, 9-9, and 9-10 have mild settings that
remove mainly the higher frequencies and still keep the important harmonics for many
instruments.
 123

10.0 78 Electric Noise Removal Methods

This chapter covers the details of removing the noise from 78 RPM records that used
electrical recording methods to amplify the music energy. The term 78 Electric will also refer
to these records.

10.1 Important Note about 78 RPM EQ Values

The DCart10 software can remove one specific type of record EQ, namely RIAA, while
it can apply many types; the specific EQ that you use for a 78 RPM record can only be a good
estimate at this point in the restoration process and you will not be able to remove it after you
create the Level Two file.
A good method to use is to pick the Phono Preamp American 78 setting as the default
setting. The 78 Electric Multifilter in the following text will show that default setting but if
you want to change this to a specific one for the records, a change can be performed to this
EQ filter in the Multifilter. The American 78 setting provides a turnover value of 500 Hz
verses the European 78 setting that provides a turnover value of 250 Hz.
The 78 Electric records have a large variation in turnover values and that some
experimentation is often required.
Because of the file structure with progressive noise removal if the best EQ for a 78
RPM record must be changed after you have produced a Level Two file you can always return
to the Level One Stage and modify the EQ you used to create a new Level Two file and save
yourself any manual noise removal work that you did to produce the Level One file.
The Part One Multifilter will be used with either 96 kHz samples per second or 44.1
kHz samples per second.
The creation of the Level One and Level Two music files will follow the same
workflow as described in the chapter General Noise Removal and will not be repeated in
detail here.

10.2 Level One Creation

1. Select classic view for the Diamond Cut Software.

2. Copy the desired music file from the original file folder and open the copy as a source
file.
 124

3. Check Channel Balance and Adjust if needed using Virtual Phono Preamp.
4. Run the 78 Electric Multifilter Part One.
5. If needed, increase the amplitude of the destination file by using the CD-Prep options
on the destination file.
6. If many noise events remain slowly increase the EzImpulse settings to remove more
noise while checking that music has not been removed.
7. Perform the Sight & Sound manual noise removal method for any remaining noise
impulses in the source file.
8. When finished, store the source file as a Level One music file.

10.3 Level Two Creation

1. Copy the new Level One music file and open the copy as a source file.
2. Run the 78 Electric Multifilter Part One and make the destination into a source file
while changing, if needed, the EQ value in either the Paragraphic or Phono Preamp.
3. Apply a Change Sample Rate operation if needed from the initial 96 kHz to 44.1
kHz to the source file. Make the result a new source file.
4. Normalize the source file to have a maximum amplitude of 0dB.
5. Apply the 78 Electric Multifilter Part Two to the source file make the resulting
destination file the new source. If continuous noise remains, carefully increase the
CNF settings while checking that desired music is not removed.
6. Trim away the remaining noise before and after the music from the source file. If
desired, add extra fade down near the end of the music.
7. When finished, store the resulting file as a new Level Two music file.

After the Level Two music file has been created, you can choose to copy the file into
the restoration music file location, add one second of silence to the beginning and end, or
continue to the Advance Section.

10.4 Details of the 78 Electric Multifilter Part One

The 78 Electric Multifilter Part One is shown in figure 10-1:

 125

Figure 10-1 78 Electric Part One Multifilter

The filter is composed of seven parts that are detailed below with their settings. The
first filter helps to shape the noise spikes so that the EzImpulse filters can recover the music
in the noise. A progression of Impulse Filters helps to find and remove noise.

Figure 10-2 DeClipper (1) 78 Electric Multifilter Part One

126

Figure 10-3 EzImpulse Filter (2) 78 Electric Multifilter Part One

Figure 10-4 EzImpulse Filter (3) 78 Electric Multifilter Part One

 127

Figure 10-5 EzImpulse Filter (4) 78 Electric Multifilter Part One

Note that the various settings are on the low side and the Solo/Brass option is on. The
Crackle setting is at the lowest value as the filter seems to detect brass music as noise with
any setting.

Figure 10-6 File Conversion Filter (5) 78 Electric Multifilter Part One

Note that the two separate channels are converted to a monaural file at this point. When
the two stereo files are converted to a monaural file the original stereo 45-degree motion that
contains horizontal and vertical information can be processed to remove either type. When
 128

the two channels are added together the vertical noise components are removed and the
horizontal music remains. The reason that the file conversion is performed after the initial
noise is removed, is that the shape of the noise is greatly modified after the two channels are
combined and the filters seem to work better when they see the original shape of the noise as
it directly comes from the record rather than after it has been converted to a monaural signal.
The next filter contains the conversion from a flat recording to having a proper EQ
applied. If you want to use a different EQ filter then shown this would be the place to use it
in figure 10-7, by either changing the settings in the Phono Preamp or removing the Phono
Preamp Filter and replacing it with the desired setting in the Paragraphic EQ filter.

Figure 10-7 Phono Preamp Filter (6) 78 Electric Multifilter Part One

The last filter limits the overall bandwidth of the music. The limits are very broad at
this point since additional limiting will be performed later during Part two. Some of the early
records have frequencies much higher than often stated and a lower value may needlessly
remove music. This filter is in figure 10-8:
 129

Figure 10-8 Bandpass Filter (7) Multifilter Part One

10.5 Details of the 78 Electric Multifilter Part Two

The 78 Electric Multifilter Part Two is shown in figure 10-5:

Figure 10-8 78 Electric Multifilter Part Two

The multifilter part two contains three CNF sections. The details of the CNFs are shown
in figure 10-9,10-10, and 10-11:
130

Figure 10-9 CNF (1) 78 Electric Multifilter Part Two

131

Figure 10-10 CNF (2) 78 Electric Multifilter Part Two

 132

Figure 10-11 CNF (3) 78 Electric Multifigure Part Two

This completes the noise removal process for many 78 Electric Records that were in
average condition. The CNF filter shown in figure 10-9 has fairly mild settings that remove
mainly the higher frequencies and still keep the important harmonics for many instruments.
 133

11.0 Diamond Disc Noise Removal Methods

This chapter will cover the details of removing the noise from Edison Diamond Disc
records. This method will be used for both Acoustic and Electric recorded records. For
Electric recorded records, frequency modification can be used as shown in the Enhancement
Chapter.
For the Diamond Disc Acoustic Records one Multifilter for Part one will be used one
for either 44.1 kHz or 96 kHz sampling rate. The general design of the Multifilter will have
the two channels combined prior to the noise removal as this method works best for vertical
recording.
The creation of the Level One and Level Two music files will follow the same
workflow as described in the chapter General Noise Removal and will not be repeated in
detail here. An overall workflow is:

11.1 Level One Creation

1. Select classic view for the Diamond Cut Software.

2. Copy the desired music file from the original file folder and open the copy as a source
file.
3. Check Channel Balance and Adjust if needed using Virtual Phono Preamp.
4. Run the Diamond Disc Acoustic Multifilter Part One.
5. If needed, increase the amplitude of the destination file by using the CD-Prep options
on the destination file.
6. If many noise events remain slowly increase the EzImpulse settings to remove more
noise while checking that music has not been removed.
7. Perform the Sight & Sound manual noise removal method for any remaining noise
impulses in the source file.
8. When finished, store the source file as a Level One music file.
 134

11.2 Level Two Creation

1. Copy the new Level One music file and open the copy as a source file.
2. Run the Diamond Disc Multifilter Part One and make the destination into a new source
file.
3. Apply a Change Sample Rate operation if needed from the initial 96 kHz to 44.1 kHz
to the source file. Make the result a new source file
4. Normalize the source file to have the maximum amplitude a value of 0dB.
5. Apply the Diamond Disc Acoustic Multifilter Part Two to the source file, and make the
resulting destination file the new source. If continuous noise remains, carefully increase
the CNF settings while checking that desired music is not removed.
6. Trim away the remaining noise before and after the music from the source file. If
desired, add extra fade down near the end of the music.
7. When finished, store the resulting file as a new Level Two music file.

After the Level Two music file has been created, you can choose to copy the file into
the restoration music file location, add one second of silence to the beginning and end, or
continue to the enhancement chapter.

11.3 Details of the Diamond Disc Multifilter Part One.

The Diamond Disc Part One is shown in figure 11-1:

 135

Figure 11-1 Diamond Disc Multifilter Part One

The filter is composed of six parts that are detailed below with their settings. The first
filter helps to shape the noise spikes so that the EzImpulse filters can recover the music in the
noise. A progression of Impulse Filters helps to find and remove noise.

Figure 11-2 DeClipper (1) Multifilter Part One

 136

Figure 11-3 File Conversion (2) Diamond Disc Multifilter Part One

Note that the two separate channels are converted to a monaural file at this point. When
the two stereo files are converted to a monaural file the original stereo 45-degree motion that
contains horizontal and vertical information can be processed to remove either type. When
the right channel is subtracted from the left channel the vertical component remains and the
horizontal is removed. The reason that the file conversion is performed after the initial noise
is removed is that the shape of the noise is greatly modified after the two channels are
combined and the filters seem to work better when they see the original shape of the noise as
it directly comes from the record, rather than after it has been converted to a monaural signal.
137

Figure 11-4 EzImpulse Filter (3) Diamond Disc Multifilter Part One

Figure 11-5 EzImpulse Filter (4) Diamond Disc Multifilter Part One
 138

Figure 11-6 EzImpulse Filter (5) Diamond Disc Multifilter Part One

Note that the various settings are on the low side and the Solo/Brass option is on.

Figure 11-7 Bandpass (6) Diamond Disc Multifilter Part One

The last filter limits the overall bandwidth of the music. The limits are very broad at
this point since additional limiting will be performed during Part two. Some of the early
records have frequencies much higher than often stated and a lower value may needlessly
remove music.
 139

11.4 Details of the Diamond Disc Multifilter Part Two

The Diamond Disc Part Two is shown in figure 11-8 and is the same for either 96 kHz
or 44.1 kHz sample rate since all Level One files are converted to 44.1 kHz:

Figure 11-8 Diamond Disc Part Two

The multifilter part two contains three CNF sections. The use of three CNF sections
helps to remove some of the high frequency noise that the acoustic records have versus the
electric recordings. The recording engineers did not remove some of the high and low
frequency noise since the playback equipment at the time had a very limited response. With
the current playback systems today extra music frequencies can be heard and using three CNF
sections helps to bring out the hidden music. Refer to figures 11-9,11-10, and 11-11 for the
details.
140

Figure 11-9 CNF (1) Diamond Disc Multifilter Part two

141

Figure 11-10 CNF (2) Diamond Disc Multifilter Part Two

 142

Figure 11-11 CNF (3) Diamond Disc Multifilter Part Two

This completes the noise removal process for many Diamond Disc records that were in
average condition. Because of Low frequency Noise most of these Records will need extra
filtering in the Advanced Section for these Records.
 143

12.0 LP/45 Noise Removal Methods

This chapter covers the details of removing noise from LP (Long Playing) and 45
records. These records were introduced around 1950 and originally produced with a monaural
recording. After 1956 stereo recordings were gradual introduced. Noise removal for the Vinyl
LP/45 records is difficult for the following reasons:

1. The type of noise that occurs with the vinyl material has a shape that can be difficult
for the impulse filters to find without using aggressive settings which remove some of
the music.
2. The surface noise of the vinyl records is low so that the quiet music passages allow the
impulse noise to be heard. This contrasts with the surface noise of the 78 RPM records
which tend to cover up some impulse noise.

12.1 LP/45 Monaural or Stereo Records?

The first LP/45 records used single channel (monaural) recording technology. The
stereo concept was later introduced in 1956 by one record producer and gradually became the
standard for all records. Because the change to stereo was gradual and backwards compatible
with monaural the record companies would sometimes produce a record that was labeled as
monaural but could be a stereo record (since the record companies would manufacture one
type of record for both markets). Some LP records that were labeled as monaural, were
actually stereo!
For records made prior to the introduction of the stereo LP/45s, the conversion to
horizontal movement to recover the music can also remove groove noise since the (L+R)
operation (to create monaural) tends to cancel common channel noise. Because the early LP
monaural records would only have horizontal movement for music the first step would be to
use the (L+R) file conversion however, this is not the best choice unless you have clear
information that the record is really a monaural recording. Since many records do have a
stereo presence even though the label may say monaural, the file conversion (L+R) is not
recommended, and the music file will be processed in all cases as a stereo recording. Use
extra care to create a monaural recording from LP and 45 RPM records.
 144

12.2 LP Stylus Size

There is another important factor about these early LPs regarding the stylus size. The
first LPs used a stylus with a radius of 1.0mils for the monaural recording. The changeover
to stereo reduced the stylus size to 0.7mils and using a 0.7mils stylus in a 1.0mils application
can introduce extra noise. However, the 1.0mils stylus in a stereo groove made for 0.7mils
can cause damage to the groove. Therefore, the best approach is to use the current stereo
stylus on all LP records. Along with the Stylus radius change other changes to the shape
occurred (conical to elliptical) and it was not always clear when these changes occurred.

12.3 Modification Possible to RIAA Curve for Normal and Worn Condition

The LP records used the standard RIAA curve when they were made, and the Diamond
Cut Software Virtual Phono Preamp would be the correct one to use for all LP and 45 RPM
records during the Level Two file creation. However, many of these records can be improved
if the Paragraphic EQ software tool (RIAA Curve or Improved) is used instead of the Virtual
Phono Preamp. The Paragraphic EQ RIAA version has slightly less low-end boost than the
Virtual Phono Preamp and a somewhat greater high-end boost. The reason for this difference
in the frequency curve is related to the specific method to implement the RIAA curve within
the Virtual Phono Preamp software and the Paragraphic filter. The Virtual Phono Preamp is
an exact rendering of the curve while the Paragraphic approximates the curve.
The reason that the Paragraphic RIAA can improve the sound over using the exact
RIAA curve can be due to two possible reasons. First, as records wear the high frequency
rapid groove motions tend to degrade greater than the lower frequency slow motions.
Secondly, the original recording may have been mastered on equipment that did not meet the
RIAA curve. The Paragraphic EQ provides some high frequency improvement that can
compensate for the high-end loss.
The best judge of which RIAA software filter to use is made while you listen. You can
try either on the Level One creation when you are listening and removing noise. Then you
can pick a specific one to use for the Level Two creation. Later, if you want you can always
return to the Level One File and pick the other RIAA software filter to make a new Level
Two file. It is best to not use the reverse feature in the different filters as a method to undo
your work on a Level Two file. Just re-do the Multifilter Part One with the other RIAA filter.
The Level One filters with different EQ settings will be shown as two separate Part
One Multifilters. To keep track of which EQ is used, the phrase Normal will be used for the
 145

regular Virtual Phono Preamp with the exact RIAA curve and the phrase Worn will be used
when the other RIAA curves are used. The choice will depend on the condition of the song
and your preference.

12.4 Sample Rate for LP & 45RPM Records

Because the EzImpulse filters need all the help they can get for Vinyl music, 96 kHz
sampling filters should be used. If you do not have the ability to transcribe the music at this
rate, then 44.1 kHz can be made to work, but the results will need more manual noise removal
and lower settings on the EzImpulse filters than as shown.
The creation of the Level One and Level Two music files will follow the same
workflow as described in the chapter General Noise Removal and will not be repeated in
detail here. An overall workflow is:

12.5 Level One Creation

1. Select classic view for the Diamond Cut Software.

2. Copy the desired music file from the original file folder and open the copy as a source
file.
3. Check Channel Balance and Adjust if needed using Virtual Phono Preamp.
4. Run the LP 45 Part One Multifilters selecting either Normal or Worn. If needed,
increase the amplitude of the destination by using the CD-Prep gain increase on the
destination file.
5. If many noise events remain slowly increase the EzImpulse settings to remove more
noise while checking that music has not been removed.
6. Perform the Sight & Sound manual noise removal method for any remaining noise
impulses in the source file.
7. When finished, store the source file as a Level One music file.

12.6 Level Two Creation

1. Copy the new Level One music file and open the copy as a source file.
2. Run the LP 45 Part One Multifilter (with normal or worn as previously selected)
and make the destination into a source file.
 146

3. Apply a Change Sample Rate operation, if needed, from the initial 96 kHz to 44.1
kHz to the source file. Make the result a new source file.
4. Normalize the source file to have a maximum amplitude of 0dB.
5. Apply the LP 45 Part Two Multifilter to the source file and make the resulting
destination file the new source.
6. Trim away the remaining noise before and after the music from the source file. If
desired, add extra fade down near the end of the music.
7. When finished, store the resulting file as a new Level Two music file.

After the Level Two music file has been created you can trim the start and end then
copy the file into the restoration music file location, add one second of silence to the beginning
and end, or continue to the enhancement chapter.

12.7 Details of the LP 45 Multifilter Part One Normal

The LP 45 Normal Multifilter Part One is shown in figure 12-1. In this figure, the
Virtual Phone Preamp is used. A progression of Impulse files helps to find and remove noise.

Figure 12-1 LP 45 Part One Normal Multifilter

 147

The filter is composed of four parts that are shown below with their settings. Note that
three noise impulse filters are used to help remove noise.

Figure 12-2 EzImpulse (1) Filter LP 45 Multifilter Part One

Figure 12-3 EzImpulse (2) Filter LP 45 Multifilter Part One

 148

Figure 12-4 EzImpulse (3) Filter LP 45 Multifilter Part One

Figure 12-5 Phono Preamp (4) LP 45 Multifilter Part One

The rumble filter has been turned on to help with low frequency noise.

12.8 Details of the LP 45 Multifilter Part One Worn

 149

The LP 45 Worn Multifilter Part One is shown in figure 12-6.

Figure 12-6 LP 45 Part One Worn Multifilter

The filter is composed of five parts that are shown below with their settings. A
progression of Impulse files helps to find and remove noise.

Figure 12-7 High Pass (3) LP 45 Multifilter Part One Worn

 150

This filter removes any rumble from entering the filters since the EQ stage does not
have a rumble filter (Low Freq Shelf if used still passes Rumble).

Figure 12-8 EzImpulse (2) LP 45 Multifilter Part One Worn

Figure 12-9 EzImpulse (3) LP 45 Multifilter Part One

 151

Figure 12-10 EzImpulse (4) LP 45 Multifilter Part One Worn

Figure 12-11 Paragraphic (5) LP 45 Multifilter Part One Worn

The Curve in this example used the RIAA EQ curve or the RIAA Phono Equalization
Curve – Improved Performance.
 152

12.9 Details of the LP 45 Multifilter Part Two

The LP 45 Multifilter Part Two is shown in figure 12-17:

Figure 12-12 LP 45 Multifilter Part Two

The multifilter part two contains one CNF section. The details of the CNF are shown
in figure 12-13:
153

Figure 12-13 CNF (1) LP 45 Multifilter part Two

This completes the LP and 45 noise removal section.

 154

Advanced Enhancement Methods & Tune Library

 155

13.0 General Concepts for the Advanced Methods

The previous Basic Restoration Section concludes with music from records that are
largely devoid of noise and Labeled as Level Two Results.
This chapter describes some general concepts to enhance the recording from your
previous restoration Level Two work. Chapters will address specific methods for different
types of recordings. A special chapter describes new methods developed for restoring the
Acoustical Recorded Records back to their original studio sound.
The last chapter describes the use of a Music Library that is available in the Diamond
Cut Software called DC Tunes. This feature is a powerful Data Base with many useful ways
of storing and finding the results of your music restoration work.

13.1 Starting Condition of Files

All Enhancement Work starts with a Level Two file that represents the best noise
removal and equalization that you could achieve for the record.

13.2 Reasons for Enhancement of the Music

The previous work performing restoration was devoted to returning the music to the
condition that the recording producer wanted for the record. The best effort that could be done
is found in the Level Two music file. For many music files this will be the music that you
will place into the final Restored location. The use of the word Enhancement will always
mean that some type of modification to the frequency content of the original desired recording
will be made. Therefore, the use of RIAA or 78 EQ by themselves is not an enhancement as
it was required for the correct playback of the music.
Having a precise definition of Enhancement for your music work will help you
achieve your goal in making the music sound the way you want.
 156

There are four types of Enhancement:

1. Modification to the recording to remove various distortions that the producer left in the
recording due to masking of the distortion from the state of the art of playback
technology at that time.
2. Modifications to the recording signal that were placed on the record due to the limited
state of the art for both recording and playback technologies.
3. Increase or decrease in the frequency range of the music to counter-act remaining high
order distortion in the music.
4. Modifications to the recording frequencies that you believe will improve the sound of
the music for their own taste in music.
5. Generation of new music frequencies that were missing or very much reduced from the
original studio recording during the Acoustical Recording Era.

13.3 Details for the Types of Enhancements

13.3.1 First Type of Enhancement

The first enhancement can remove certain frequencies that were recorded in the music
but were not a concern to the producer at the time since these frequencies could not be
reproduced during the playback when using the equipment at the time. Consider for example
an acoustic recording that had significant frequencies present above or below the limited
playback range from about 300 Hz to 3000 Hz. These noise frequencies can now be heard
with today’s audio equipment and should be removed. When the music was recorded the
audio engineers, even if they knew that extraneous noise was present, did not bother to avoid
this noise as long as the playback equipment couldn’t reproduce the noise.
An acoustic recording that may have recorded low bass frequencies in the same
location as turntable rumble noise. If these low bass music frequencies are simply removed
along with the turntable rumble noise, then a possible improvement to the acoustic recording
has been lost. As the enhancement tools are used, a balance will often be needed between
removing real noise and real music.
The high frequencies contain surface noise from the material used in these early records
and can easily dominate the desired music. Many acoustic and even later electric playback
phonographs did not reproduce frequencies above 5,000 Hz. Today we hear the scratchy
sound and it degrades the music.
 157

The Edison Diamond Disc Records are an example of this type of recording that can
be enhanced by removal of low frequency rumble that occurs in many of these records when
played back on today’s Turntables.

13.3.2 Second Type of Enhancement

The second enhancement will use your own hearing to determine the amount of music
signal modification that maybe needed. Some examples will help to explain the problems and
solutions. Early electric music customers were preconditioned to expect a certain sound when
a record was played back due to their experience hearing music on the new radio technology.
The use of radios predated the introduction of electric recorded records and electric playback
(many early radios used battery power as widespread home power was not present); when a
new electric recording of a song was first heard the expectation was to hear a similar range of
music frequencies as heard on the new radios. Later into the early 1930s, many listeners of
records were hearing records with deep bass due to the introduction of jukeboxes in public
spaces. The record producers would often boost highs and lows to counteract the playback
systems own frequency modifications to give the sound they wanted you to hear. Today with
our hi-fi audio systems that can respond from 20-20,000 Hz the same record that sounded
good in 1932 will often sound wrong with a modern system. Converting the recording to
sound natural with today’s audio systems will be a second type of enhancement.
In the Handbook sections that describe equalization the use of Pre-Emphasis is a
method to extend the upper frequency range of the music by reducing the effect of surface
noise on the music. This concept was first discussed in the 1925 paper by the Western Electric
Laboratory regarding electric recording, however it first gained commercial use much later
in FM (Frequency Modulation) radio. This boosting of the higher frequencies was applied to
some of the 78 RPM records; however, this practice was not uniform with the record
companies and was not well documented.
For the 78 Electric Level Two results the equalization consists of using a single
turnover value and does not decrease the frequencies above the turnover starting value.
Therefore, there will be some records that had an increase in the higher frequencies and this
extra boost during recording may need to be reduced for playback with today’s audio systems.
Another type of record that fits in this section is the Edison Diamond Disc Electric
recorded records. These records were developed by Edison independent from the 78 RPM
Electric records and have their own unique EQ. Your own hearing will be the best method to
adjust the finished restored product.
 158

13.3.3 Third Type of Enhancement

The general subject on the best range of frequencies to reproduce music was studied in
detail from the mid 1920’s through the 1950’s when many pioneering scientists started in-
depth studies into how we hear speech and music. The cost of many audio devices from radio
receivers to broadcasting stations are price sensitive to the range of frequencies that are
reproduced which in turn helped to sponsor this research. The Audio Engineering work
performed by Harry Olson in his book Elements of Acoustical Engineering is one of many
excellent sources that details important findings regarding the relationships between the
frequency range of the music, the distortion in the music, and the perception by humans on
the quality of the music. A brief listing of some of the results from the study of how we hear
music will follow. For our discussion, distortion in the music will be defined as frequency
modifications to the notes in the music. Musical instruments create harmonic frequencies
from the base fundamental note and any distortion added to these higher order harmonics will
degrade the sound. An example of frequency modifications would be a tearing sound in a
brass instrument. Some of the key findings were:

1. If the music has very little distortion, then a wide frequency range during playback will
enhance the quality of the music as heard by a listener.
2. For music with some noticeable distortion, a reduced frequency range will often help
to reduce the effect of the distortion on the music as heard during the playback to the
listener. The reduced range is for both the low and high frequencies.
3. The range of music should be such that the arithmetic mean of the frequency range is
centered around 800 Hz in order to produce a balanced overall sound. Therefore, if you
reduce the upper frequency limit you need to reduce the lowest frequency by a
corresponding amount to keep the mean around 800 Hz. Another similar method is to
have the product of the lowest and highest frequency be approximately 500,000 (Hz
Squared) for a balanced sound.

13.3.4 Fourth Type of Enhancement

The fourth type of enhancement provides an opportunity for you to perform some music
creativity. The Diamond Cut Software provides methods to modify the sound from the
recording to satisfy your desires independent of the original studio sound. For example, a
pseudo-stereo effect from the original monaural can be achieved if desired through time
 159

offsets and selective right and left channel filtering. Extra reverberation or echo can be
introduced if want to add a lively sound to the music. This handbook provides some examples
for these methods.

13.3.5 Fifth Type of Enhancement

The fifth type of enhancement is a new area of research that I have developed. Many
of the early Acoustical Records used musical instruments that produced frequencies over
which their total range was not recorded. The lack of recording low frequencies, that were
often the fundamental tones, meant that the music had a tinny sound that was not what was
heard in the recording studio. Many of the higher harmonics to the instrument’s sound did get
recorded on these early records which opens the possibility of a new technique to add back to
the music; the missing fundamental tone.
While this concept appears to be difficult the idea of replacing the missing fundamental
note is performed by our brain when we hear certain types of music. The general study of
Psychoacoustics describes this and other aspects to hearing and should be consulted for more
information. The encoding of music to the various standards (MP3) takes advantage of
psychoacoustics and removes some of the music frequencies to minimize the file size with
the expectation that the human brain will create the missing values.

13.4 Examples of Enhancement Multifilters

Later chapters will provide enhancement examples that include a combination of types
one, two, three, four, and five that are tailored for various types of recordings. The specific
settings shown in these Multifilters are starting points for your enhancement work. The final
setting that you use will often be unique for each record. Although the examples are specific
for a certain type of record, they can also be applied to other types of records.

13.5 Saving the Results of Your Enhancement

The music that you use for the enhancement work comes from the Level Two location.
The results of the enhancement should be placed in the Restored File location. You will want
to keep the original Level Two file before the enhancement operation so that you can perform
more enhancement on the Level Two music as time goes on.
 160

14.0 Enhancement First Method Example (Rumble & Hiss)

This example will use an Edison Diamond Disc Acoustic Record, but the methods can
be used on other Acoustically recorded songs.
The Edison Diamond Disc Acoustical Records have many songs that contain extra low
and high frequency distortion. Because these noise frequencies are not always present and
there is some music present with the noise the various filters used in the Level Two restoration
section were set to wide ranges. When rumble and hiss are present more removal is needed
from the Level Two results. First, we will remove low frequency noise caused rumble.
Rumble in the Acoustical Recorded Diamond Discs was caused by manufacturing
defects in the surface that allow the stylus to have a vertical motion that was not due to the
music. Examination of these disc surface shows bumps and depressions that usually cause a
thump sound once per revolution. A close examination of this type of noise has shown that
most of the frequencies in the thump sound are below 150 Hz. When the same record that has
a thump is played on a period Edison Phonograph the noise is not heard since the Edison
Phonographs low end is around 200 Hz. A series of figures will show this in more detail and
the solution to the noise will be to remove all frequencies below 150 Hz for a noisily record.
The following is a frequency response of an Acoustic Recording as heard from a period
Edison Phonograph (the noise spike at 120 Hz is related to an environmental sound during
recording and not from the record).
The Song that was played was The Charleston by the Golden Gate Orchestra. Refer to
figure 14-1:
 161

Figure 14-1 Charleston as heard from an Edison Record Player

The frequency output is about 200-3000 Hz and no low-end distortion or thumps, or

hiss were heard. The Level Two frequency response of this song is in figure 14-2:
 162

Figure 14-2 Charleston from Level 2 Restoration

Notice the large amount of low frequency energy below 150 Hz that appears to be noise
rather than music. The high end shows the result of the Level 2 filtering with the steep high-
end drop-off. Both Low and High frequencies will need some modification.

14.1 Rumble Removal Acoustic Diamond Discs

The frequencies from 200 Hz and lower contain noise with a very small amount of
music that is masked or covered over by the rumble. The frequencies and noise that will
remain in this area after filtering are a compromise between hearing the rumble and keeping
some music. After many tests on these Diamond Disc Records, a Continuous Noise Filter at
150 Hz Cut-Off was found to work well. The Multifilter for this rumble is shown in figure
14-3:

Figure 14-3 Multifilter for Rumble Removal

The details of the continuous Noise Filter are in figure 14-4:

163

Figure 14-4 CNF (1) for Multifilter

The result of using this multifilter on the Charleston Level Two Song is in figure 14-5:
 164

Figure 14-5 Charleston after De-Thump Filter

Compare 14-5 to 14-2 to see how the low frequency noise is removed. When this song
is played back the rumble is gone with a much better low-end sound. The high-end noise will
now be removed.

14.2 Surface Noise Removal and Final Cleanup Acoustic Diamond Discs

The high-end frequencies will often contain music and record surface noise. When the
high-end surface noise frequencies are simply removed by a sharp filter the music can have a
somewhat dull or lack of sparkle to the sound. It is easy to cut off music frequencies along
with surface noise at this point so an alternative to a low-pass filter is to boost the frequencies
somewhat just before the high frequency cut-off and then use a somewhat gradual CNF to
limit the scratch sounds. By slightly boosting the higher frequencies just before the cut-off
you are helping your hearing to replace some of the lost music to the surface noise. Note that
this method will need adjustment for each record as the surface noise varies from record to
record.
 165

Figure 14-6 shows the Multifilter used with the name Diamond Disc Acoustic Cleanup
Multifilter. Since this filter is the last operation that is used on the file (after the De-Thump
Removal) some low frequency boosting is also present. However, the low frequencies are
missing from much of the music and many of them cannot be boosted much at this point.

Figure 14-6 Diamond Disc Acoustic Cleanup Multifilter

Figure 14-7 shows the first block which is used to slightly lower the amplitude so that
any later boosting will not saturate the music file:
 166

Figure 14-7 Gain Block (1) Diamond Disc Acoustic Cleanup Multifilter

The next filter boosts some of the frequencies in the music. Refer to figure 14-8:

Figure 14-8 Frequency Boosting EQ (2) Diamond Disc Acoustic Cleanup Multifilter

The last filter cleans up surface noise that can be heard. Refer to figure 14-9:
 167

Figure 14-9 CNF (3) Diamond Disc Acoustic Cleanup Multifilter

This concludes the section on the First type of Music Enhancement for a Diamond Disc
Acoustic Record. The same general approach can be used for 78 RPM Acoustic Records or
other early Acoustic recordings (Cylinder Records).
 168

15.0 Enhancement Second Method for Various Record Types

15.1 78 RPM & LP\45 Records

This enhancement shares some of the techniques used for EQ selection for both 78
RPM Electric and LP\45 RPM records. When the Level Two recordings were made for these
records, you had an opportunity to use the standard EQ values or a modification. For example,
the Worn EQ for LP or different turnover values for 78 RPM records. A further modification
can be made using the 10 Band Graphic EQ filter under the Filter Tab. A 20 Band version is
another software option but having ten ranges is very adequate and is quite a change from
tradition Bass and Treble controls. A picture of the filter to reduce high frequency from added
pre-emphasis in the recording and slightly boost the low end is in figure 15-1:

Figure 15-1 Type Two Enhancement Example

The settings used in this filter will be set to a value that sounds correct to your hearing.
Note that instead of decreasing the higher frequencies, you could boost them for dull sounding
LP/45 Records.
 169

15.2 Diamond Disc Electric Enhancement

The Edison Diamond Discs that used electric recording benefited from a much wider
frequency range than the Diamond Disc Acoustic recorded records. The resulting
enhancement multifilter has settings that will reflect this wide range.
The Edison Electric Recorded Diamond Disc technology used a different recording
method than the new Western Electric technique. The Western Electric recording method was
specific for the lateral or horizontal stylus motion which contrasts with the use of a vertical
motion for the Edison Diamond Discs records. Little information is known about the technical
details that Edison used for these records. The multifilter here was found using trial and error
methods.
During the brief time that Edison produced these records the quality of the recordings
varied with some having a strong bass while others were thin sounding. Many of the records
contained large amounts of low frequency rumble that was close in frequency to the desired
music frequencies since the phonographs used for playback were only mechanical and could
not reproduce these low frequencies. Edison did develop electrical playback phonographs
with EQ at the very end of the 1920’s (needle type records) but shut down all music
production shortly afterwards. Because of the large variation in sound each record will need
unique settings.

15.2.1 Diamond Disc Electric Enhancement Method

The Edison Electric Recordings reproduce new low and high frequencies from the
previous Acoustic Recordings so the method described in 14.0 will be modified for use with
these records.
The first step before enhancement is to remove the rumble or low frequency noise
sounds from the songs that were present in the original recordings. The Edison music
engineers knew that the Edison phonographs at the time could not reproduce many low
frequencies below 200 Hz so that the presence of low frequency noise was not a problem for
them at this time in the 1920’s. However, since we can reproduce the inherent noise today
this noise is a problem for us. An additional complication comes from the fact that music has
many frequencies at the same location as these rumble frequencies so that a simple removal,
as was done with the Acoustic Diamond Disc records of all the music frequencies below 150
Hz, would degrade the quality of the song.
 170

15.2.2 Check for Strong Discrete Rumble Frequencies

The first step is to play back a Level Two recording and use the spectrum analyzer to
check for any significant low-end rumble frequencies. The following figure 15-2 shows noise
at 87 Hz and 123 Hz that was found on this Level 2 recording. This noise was reduced some
in the previous Part Two Multifilter for the Electric Diamond Disc records. However, there
is a good amount of music information in the same low-end area so we cannot always remove
these rumble frequencies during the Level Two file creation.

Figure 15-2 Rumble noises.

The marker is located at 123 Hz and the other peak at 87 Hz can be seen. The noise at
123 Hz could have been introduced from the initial transfer made at 78 RPM since common
120 Hz noise when multiplied by the speed increase (78.26 RPM to 80 RPM) is about 123
Hz. The noise at 87 Hz is from the original manufacturing process. Notice that the general
frequencies below 100 Hz have quite a bit of energy that is not from the music.
If we did not remove the rumble noise, then the improved sound from the enhancement
process would have a strong steady drone in the song. These rumble frequencies can be
removed by using individual notch filters on the Level Two file for any noise that was found.
 171

The Edison Diamond Disc Records at the time that electric recording was used have a
wide variation in the noise. It is equally possible to find records that are very clean as to find
records that have noise.
Figure 15-3 shows the notch filter for 123 Hz. The idea is to remove the noise and not
the music that is close by. The notch filter is just the right tool for this.

Figure 15-3 123 Hz Notch Filter

Figure 15-4 Shows a notch for 87 Hz:

Figure 15-4 87 Hz Notch Filter

The spectrum of the same Level Two music after the application of the two Notch
filters is in figure 15-5:
 172

Figure 15-5 Notch Result

The rumble at the previous frequencies has been removed. Any low and moderate level
rumble frequencies will be further removed in the enhancement multifilter.

15.2.3 Electric Diamond Disc Enhancement Multifilter

Now that strong rumble discrete frequencies have been removed, a Multifilter can be
applied to the music. Since a specific EQ curve for these records is not known, a Shelf Type
Bass and Treble control can be effectively used. The use of the phrase shelf means that the
amount of boost continues to a final fixed value or shelf. The tone controls in the Phono
Preamp implement this type of filter. Refer to figure 15-6 for this multifilter:
173

Figure 15-6 DD Electric Enhancement Multifilter

The filter details are in figures 15-7 to 15-10:

Figure 15-7 CNF (1) DD Electric Enhancement Filter

174

Figure 15-8 Filter (2) DD Electric Enhancement Filter

Figure 15-9 Filter (3) DD Electric Enhancement Filter

 175

Figure 15-10 CNF (4) DD Electric Enhancement Filter

These last two filters will remove any left-over background or continuous noise. A
unique setting for your recording may be needed, but these settings are a good starting point.
The values for all filters can be optimized for each record due to the large variation in
these Electric Recorded Diamond Discs.
 176

16.0 Enhancement Third Method

This method builds on the work of Audio Scientists research into the best frequency
range to be used for the playback of music. One excellent source is contained in the book
Elements of Acoustical Engineering by Harry F. Olson. Internet searches for his book will
find both Used Books and pdf versions available and should be part of your reference library.
The key points the were discovered were First that the presence of high-order
distortion products in the higher frequencies will causes the music to sound un-natural and
irritating to the human ear to the point where the music will improve by lowering the upper
frequency limit to the song. For our restoration and enhancement work our playback
equipment (modern hi-fi & large speakers) should not introduce any additional distortion
products. Therefore, any distortion that we hear in the Level Two music is present in the
recording and removal at this point cannot be performed. What is possible is to reduce the
high end of the music as much as needed to remove the distortion sound while keeping a
natural sound to the music. If a change to the high-end recording is needed, then a further
improvement can be made by the Second key point found.
The second key point was that the playback of music sounded best to the human ear if
either the product of the range of music frequencies was around 500,000 or the mean of the
music’s frequencies was 800. The overall result should have a balance to the music with the
lows and highs on either side of a scale with the balance point around 800 Hz. Keep in mind
that early music, while having some low frequencies, doesn’t have the correct amplitude or
strength for these low frequencies due to the recording technology used at the time. Check
both sides of the frequency spectrum to create a balance point around 800 Hz between equal
amplitudes of the low and high frequencies.
These findings were developed after electrical recording had been established and
should be used with care when you have acoustical recordings for enhancement. The
acoustical recordings were often made using a reduction in the number of instruments in the
recording studio to try and make the best recording possible with the technology available.
For these types of records the balance point may be different than electrical recordings.
This enhancement can be implemented by using the CNF (Continuous Noise Filter) in
a somewhat unique manner. The music should be heard and a small adjustment made (using
the CNF) in reducing frequencies for a song with moderate distortion to improve the quality
of the song. The range of frequencies should have a mid-point around 800 Hz so that higher
frequencies are reduced along with the low values.
 177

The CNF filter is an ideal method to control the range of the frequencies that you want
to be heard by adjusting the total range via the movable frequency levels on the low and high
end. For an example refer to figures 16-1 and 16-2:

Figure 16-1 Flat CNF

 178

Figure 16-2 Frequency Range Decreased CNF

In figure 16-1, the CNF will pass all frequencies with little amplitude change since the
bar for change is low. However, figure 16-2 shows a reduced range of approximately 60-6000
Hz set by raising the level that the signal must exceed to have no effect. The other settings in
the CNF (FFT Size and others) should be adjusted for best effect. The concept demonstrated
by figure 16-2 is that an adjustable range to the music can be performed by using the CNF.
 179

17.0 Enhancement Fourth Method

This method allows for your creativity for music change from the original studio sound.
A general example for 78 Electric Records that highlights different effects in the DCart10
software is in figure 17-1. The same method can work for other types of records. You are
encouraged to modify the settings shown to achieve a good result for each song.

Figure 17-1 78 Electric Enhancement Multifilter

This multifilter is composed of many filters and effects. The order and the resulting
processing are important for this filter and by moving the locations within the multifilter the
result can be very different. The reason that the specific signal processing order is important
is due to the non-linear operation of some of these effects and the fact that one stage expects
that some specific signal processing has already occurred. You can modify the individual
settings later after gaining a good understanding of the operation of each one. The filters and
effects can be grouped into some general sections as shown:
 180

1. Noise Filter. The first filter is a continuous filter that has more aggressive settings than
the general one used in Part Two of the 78 Electric Multifilters.
2. The two effects labeled two and three provide a method to create new frequencies that
are related to those present from the music. The use of these effects can be very
powerful in restoring music frequencies that were either not recorded due to early
technology limitations or lost due to record wear.
3. Effects labeled four and five are used to expand the amplitude of certain frequencies
and then to compress some of the highest levels. While these expand and compress
effects may appear to be at odds with each other using them together is the best method
to balance the overall dynamic range of the music.
4. The last three effects, labeled six, seven, and eight, can add extra frequencies that are
related to the music frequencies present in the music. This relationship to the recorded
frequencies is different than the effects stated in locations two and three.

Each section of the multifilter, along with the current settings are shown in the
following figures 17-2 to 17-9:

Figure 17-2 CNF (1) 78 Electric Enhancement

181

Figure17- 3 Subharmonic Filter (2) 78 Electric Enhancement

Figure 17-4 Overtones (3) 78 Electric Enhancement

182

Figure 17-5 Punch & Crunch (4) 78 Electric Enhancement

Figure 17-6 Dynamics (5) 78 Electric Enhancement

183

Figure 17-7 Virtual Valve (6) 78 Electric Enhancement

Figure 17- 8 Virtual Valve (7) 78 Electric Enhancement

 184

Figure 17-9 Virtual Valve (8) 78 Electric Enhancement

17.1 Operation and Flow of the Enhancement Features

The overall music flow is from left to right within the multifilter. The music enters the
CNF (continuous noise filter) where the specific range of frequencies to be enhanced are set.
After the music has had as much noise removed without adversely affecting the quality of the
song, the signal enters stages that will add frequencies to the music. The multifilter can be
thought of as first removing noise and then adding to the music.
Both sections two and three increase the frequency content by adding frequencies to
the music that are harmonically related to the music. Help files in DCart10 can provide
additional information for these effects (2 & 3). The music frequencies are then selectively
increased in section four greater than they are in the original music (punch effect). Section
five is then used to smooth out or limit any signal levels that were boosted too much. The last
sections help to add in various harmonics related to the song in a manner that has more control
over the music than simple bass and treble tone controls would.
 185

18.0 Enhancement Method Five

18.1 Introduction to this Method

This technique was originally described in an IEEE (Institute of Electrical and

Electronics Engineers) Spectrum Magazine article (February 2017). This chapter contains an
expanded version of the method tailored to the restoration methods and software described in
this Handbook.
I have enjoyed listening to early recorded music on Period Phonographs. The sound
from these antique phonographs, while enjoyable, always lacked many frequencies due to the
technology used at the time. For many years I tried to improve the sound by boosting the low
and high frequencies using tone controls and then multi-band equalizers. These static methods
failed to improve the music but did increase the record noise! I developed this new method
specifically for Acoustical Recorded Music (Music recorded without any electrical devices)
and the results are a significant improvement in the overall balance of this new version of the
old music.
The method uses a special type of dynamic frequency modification for use on
acoustically recorded records by taking advantage of the Harmonic Relationship of Music
from Instruments.
The purpose of this modification is to bring the overall music experience back to what
was originally heard in the recording studio by placing in the song additional new frequencies,
that were present in the recording studio, but were not recorded in the original records.
In this chapter I will use the phrase Correction to emphasize that this effort will restore
the sound to what was heard in the studio given that most of the recordings are 100 years old.
There are no musicians or production engineers available to guide this work on
correction of these recordings. The best that can be done is to listen to the results of your work
and then use your musical experiences to guide the modifications to the original acoustic
recordings. The result should be to produce a song that sounds as close as possible to a natural
sound from the instruments.
While this method can work for other Acoustical Recorded Songs the Examples shown
are from acoustically recorded Edison Diamond Discs. A description of the method is in 18.2
and the implementation is in 18.3.
 186

18.2 Description of Correction Method

The method for the correction of acoustical recorded songs requires several processing
steps that will be discuss. The starting condition is the result from a Level Two operation as
previously described in the Handbook. The Acoustic Recorded Song, The Charleston, played
by the Golden Gate Orchestra on an Edison Diamond Disc from April of 1925 will be used
as an example.
When acoustic recorded music was first heard by the public, the sound from the
musicians had been modified by the original recording of the studio master and in turn by the
mechanical playback of the record used in the period phonographs. By using the results from
the transcription, using today’s audio equipment, and then removing much of the noise (Level
Two), the limitations to the Acoustic Process from the mechanical playback have been
removed. What remains in the music is the result of the original mechanical recording method
that was state of the art at this time.
At this point in the correction, the Level Two music contains three general frequency
sections. The first section is called the Low frequency area from 0 to about 300 Hz which
contains noise, some music, and the remaining harmonic frequencies from fundamental notes
that were NOT recorded. The second section is a sub-division within the Low frequency area
from 150 Hz to about 300 Hz which contains music frequencies, both fundamental and
harmonics, that had a reduction in their proper amplitude. The last section from 300 Hz up to
6000 Hz (or even higher) contains a good, faithful recording of the studio music. The
development of electrical recording was a huge improvement to the music sound, but the late
stage acoustically recorded music did a good job at capturing the frequencies from 300 Hz
and higher.
The Key to this Correction method is to create new music low frequencies below 300
Hz and removing the low frequency noise. Music frequencies from 75 Hz to 300 Hz will now
contain both new frequencies and originally recorded frequencies with the new frequencies
being the missing fundamental notes. The area from 300 Hz and lower will become the Low
frequency section which is later merged with the High frequency area above 300 Hz to
produce the final music result.
The method used to create these new fundamental frequencies uses a software
algorithm from DCart10 that takes the harmonics of musical instruments and then produces
a new corresponding lower fundamental note (that was missing) from these higher harmonics
for each instrument. Extra steps in the low frequency generation will provide the needed
transition from these new to original frequencies from 75 Hz to 300 Hz.
 187

18.2.1 First; Remove Rumble and Generate Source

The acoustically recorded songs contain a significant amount of low frequency noise
that would not be heard using a Period Phonograph. The same songs’ frequency content, when
played back on a Period Phonograph (Edison Sheraton Model S-19), is shown in figure 18-1.

Figure 18-1 Frequency Content of Charleston Song from Period Phonograph

The frequency content of the same song transcribed and restored to a Level Two File
is in figure 18-2:
 188

Figure 18-2 Level 2 Frequency Spectrum of Song the Charleston

In this first step, a Source File will be generated which is the result of the Rumble
Removal process using the First Method Enhancement described in Chapter 14 for the Low
Frequency noise (CNF set to 150 Hz). This low frequency noise can also be described as
having a Thump sound when the record is played on today’s hi-fi equipment. This cleaned
file will now become the Source File for generating new frequencies. This source file will
provide both the portion of the music that needs new low frequencies and the portion of the
music that was recorded correctly. The new Files from the source will be labeled Low and
High. The results of the De-Thumper (Low Frequency Noise Removal) Multifilter is in figure
18-3.
 189

Figure 18-3 Charleston Source File

18.2.2 Second; Generate Low Frequency Correction File

The acoustic recording process had a limited range with the lowest frequencies
recorded, but not accurately, at approximately 150 Hz. Many musical instruments have a
fundamental or base note lower than 300 Hz. Music charts and tables will show the various
frequency ranges needed for each instrument to be correctly heard. Since Middle C has a
fundamental frequency of 261.63 Hz, the fact that Acoustical Recorded songs sound tinny is
not un-expected. The understanding that instruments have harmonics that are related to the
fundamental note of each instrument presents a method for generating the missing
fundamental notes via an application of an algorithm in the Diamond Cut Production software
program described as a Sub-Harmonic Synthesizer.
The operation of this software is that when a frequency enters the algorithm, a new
frequency and the original will exit the operation. The new generated frequency will be one-
half of the original value. The software contains adjustable settings that are shown in figure
18-4:
 190

Figure 18-4 Sub-Harmonic Controls

Sub Level controls the amplitude of the new generated frequencies, Frequency
establishes the high end of the operation, Cents adds fine control, the Male Vocal
Discriminator helps to make a male singer sound natural, the Sharp Cutoff helps to limit
the filter performance and the Output Level controls the result. A sine wave of 200 Hz was
fed into this synthesizer (settings shown) and the output is in figure 18-5:

Figure 18-5 A new frequency of 100 Hz along with the original 200 Hz.
 191

The Sub-Harmonic Synthesizer will always produce frequencies that are related to the
input values in a fixed relationship of one-half (with the cents at 0). If a tuba or bass instrument
has a harmonic at 250 Hz then the original fundamental at 125 Hz (which would not be
recorded) will now be re-introduced into the music. The potential for a large increase in the
quality of the song is possible since you are not simply boosting low notes, rather you are
creating the notes that were originally present and were lost in the recording process but
present in the music studio. This is a Dynamic Process in Sharp Contrast to a Static Process
that uses fixed Levels of Boost.
During the implementation there will be additional filters added to the synthesizer to
optimize the generation and mixing of these new frequencies. The collection of components
to perform this process is called the Low Multifilter (details in the implementation section),
and the figure that follows represent the results of using this Multifilter.
To aid in seeing the results, the frequency range from 20-500 Hz will be graphed for
the Source File and resulting new Low frequencies. First, the source from 20-500 Hz is shown
in figure 18-6:

Figure 18-6 Source Spectrum 20-500 Hz from the Charleston.

 192

There are no significant frequencies below 150 Hz. Now, this Source file will be sent
though the Low Multifilter to yield the new Low Spectrum. Refer to figure 18-7:

Figure 18-7 Output of Low Multifilter 20-500 Hz for The Charleston.

At this point new frequencies along with original frequencies can be seen. From 75 Hz
to 150 Hz shows these new frequencies related to the original recorded values. The area below
75 Hz contains noise and some frequencies related to the finite nature of filters. The area from
150 Hz to 300 Hz shows a blending effect of new and present frequencies due to the finite
nature of signal processing and an additional filter after the sub-harmonic synthesizer.
The same vertical amplitude scale was used in both pictures. Next, the high frequency
file will be produced.

18.2.3 Third; Generate High Frequency Correction File

The previous Source for the Low frequencies will again be used to produce a high frequency
music file. This high file will have modifications made to its low end by using a steep high
pass filter set to 300 Hz. The frequency of 300 Hz corresponds to the approximate transition
 193

point from missing low fundamental frequencies to recording the music correctly. Note the
reduction in the lower frequencies from the source file (figure 18-6) after using this filter. The
result is in figure 18-8.

Figure 18-8 High Multifilter Result 20-500 Hz.

18.2.4 Fourth; Merging of High & Low Frequency Files

The music does not have a sharply defined point from needing, to not needing, new
low frequencies. The original acoustic recording process had a frequency range over which
the recorded frequencies slowly improved to be fully captured by the master recording.
Because of this gradual change-over in the recording performance, the selection of the settings
in the sub-harmonic generator (sharp filter not checked) along with an extra filter helps to
create this needed transition frequency range. The amplitude control used during the merging
of the new Low and High music files will also help to make the music sound natural.
Custom amplitude adjustment on the amount of mixing the Low results with the High
results will be needed that will depend on the specific music content. Too much addition of
these New/Lost frequencies can cause an un-natural sound to the music since the sub-
 194

harmonic synthesize does not have special knowledge of the song. The frequency range
graphed is from 20-500 Hz and is shown in figure 18-9 for the merged results:

Figure 18-9 High & Low Music Files 20-500 Hz

The combined result over the total audio frequency range is in figure 18-10:
 195

Figure 18-10 High & Low Music Files over total range.

Notice from left to right the smooth transition from the new Low frequencies into the
portion of the music that was recorded correctly and then out to a quite high value for an
acoustically recorded song (over 5000 Hz).

18.2.5 Fifth; Cleanup / Final Modification

After the merging of the files further frequency modification is now possible since low
and high frequencies are present that have low distortion. A slight high-end boost followed
by a steep decline will add some brilliance to the music while limiting surface noise. The low
end can be boosted because music frequencies are now present rather than just noise. The
cleanup multifilter (details in the implementation section) was used to perform the final
frequency changes shown in figure 18-11:
 196

Figure 18-11 Finished Music File.

Figure 18-11 shows the large increase in the music range after using this Acoustical
Correction Method.
Figure 18-12 shows an overlay of the finished correction with the same song as heard
on the Period Phonograph (lighter values) using the same vertical scales. The finished result
greatly expands the overall range of music from the original range heard on the period
phonograph.
 197

Figure 18-12 Final Correction and Version Heard on Period Phonograph

The result should be judged comparing the sound heard from this process rather than
just spectrum values from figure 18-11. Most people that have heard this new version of the
Charleston state that the sound has a balanced and full sound to it and enjoy hearing this new
version of an old song.
The various multifilters used do not have special intelligence about what frequencies
are created. Extra steps have been placed into the various multifilters to minimize artificial
sounds; however, users of this technique can experiment and may find that adjustments to the
described settings may help to improve the process for their music.
This technique has also been used on Acoustical Recorded 78 RPM records with
similar success. An implementation of the method using Diamond Cut Productions software
will now be shown.
 198

18.3 Implementation of the Acoustical Correction Process

The settings shown in the following implementation represent starting values that have
provided good results for many Acoustic Recorded Records. An optimum setting for each
song maybe needed since songs have unique frequencies and combination of musical notes.

18.3.1 Create A Working Copy

A copy of your Level 2 file will be used as the working file to allow you to return to
the original Level 2 work for future improvements if wanted. The extra files that will be
created from this working file should be saved with unique names such as Source, Low, High,
etc.

18.3.2 Create a Source File

The first new file will have the thumps and noise from production removed. These low
frequency noises are present on many early acoustical records. The result of using this
multifilter will be saved and renamed as the Source File. The de-thumper multifilter shown
in figure 18-13 was found to remove as much noise as possible while keeping most of the
music low frequencies. Refer to figure 18-13:

Figure 18-13 Acoustic De-Thump Multifilter

 199

The single filter in this multi-Filter is a Continuous Noise Filter. The settings are in
Figure 18-14:

Figure 18-14 CNF Settings in De-Thump Filter

18.3.3 Create A Low File

The previously created Source File will now be used to create the file that contains
both new low frequencies and original frequencies below 300 Hz. First the source file is sent
through the Low multifilter, and the resulting file is labeled Low. Refer to figure 18-15 for
the low multifilter:
 200

Figure 18-15 Acoustic Correction Low Multifilter

This filter has two components. The first creates the new frequencies and the settings
for the Subharmonic are in Figure 18-16:

Figure 18-16 Subharmonic Settings

 201

The next filter stops the generation of extra new low frequencies, though the use of a
brick wall FIR (Finite Impulse Response) filter. The settings are shown in figure 18-17:

Figure 18-17 Bick Wall Filter

18.3.4 Create A High File

Next the previously created Source File will be used to create the file that contains the
portion of the original recording that recorded correctly. The source file is sent into the high
multifilter with the resulting file labeled High. Refer to figure 18-18 for the multifilter:
 202

Figure 18-18 Acoustic Correction High Multifilter

This multifilter consists of one Bick Wall filter with settings shown in figure 18-19:

Figure 18-19 Brick Wall One Filter Settings

18.3.5 Create High & Low File

 203

After you have a separate High and Low file created from the source file, these files
are merged into one resulting High & Low file so that when added together they have a good
balance. This is performed using two DCart10 software features in a non-standard manner.
First, the new High and Low file will become a Stereo File with the Left channel High
Frequencies and the Right Channel Low Frequencies. The file split and recombine feature
under the Edit tab will be used with a new destination file named Combined. Refer to figure
18-20:

Figure 18-20 Create Stereo File

The result of this operation for an Acoustic File (Charleston is in figure 18-21:

Figure 18-21

Now that we have two separate files contained in a Stereo File (Left & Right Channel),
we can mix them back to a Monaural file and vary the amount of mixing of Low and High by
using the file conversions software under the Filter tab. Refer to figure 18-22:
 204

Figure 18-22

By using the Preview button, you can balance the amount of Low that is added to High
frequencies to make the song sound natural before you commit to a specific setting. When
you have determined a gain setting to use you then generate a merged destination monoaural
file. Giving the destination a new unique name allows you to keep the original high & low
file for a possible re-do.
Refer to figure 18-23 for a picture of the merged file containing the mixture of the High
and Low files for the previous song:

Figure 18-23 High and Low files Mixed into a New Merged File

18.3.6 Create the Final Cleanup Version

The last operation is to send the merged file through the cleanup multifilter to add some
bass and treble and to eliminate any high frequencies that are not music but just surface noise.
Refer to the Cleanup Multifilter in figure 18-24:
 205

Figure 18-24 Acoustic Cleanup Multifilter

The first part of the multifilter uses the file conversion software to reduce the overall
amplitude so that any boosting of frequencies will not saturate the signal. Figure 18-25 shows
the settings:

Figure 18-25 Stage 1 gain change in the multifilter

 206

The next section modifies the low and high frequencies by boosting them now that
music frequencies are present. Refer to figure 18-26 for the details:

Figure 18-26 Stage 2 frequency changes in the multifilter

The last filter removes some left-over noise that does not contain music. Refer to figure
18-27:
 207

Figure 18-27 Stage 3 CNF filter in the Cleanup Multifilter

18.4 Conclusion

The purpose of this newly developed method is to restore the music from the record to
what was originally heard in the music studio when the recording was made. This is a unique
method I have developed and has been long been a pursuit of mine to bring more life and
music out of these unique records.
These records preserve an important time in our history of music. This modification to
the music should make it sound as if you were in the studio when originally recorded and in
turn appeal to a wide audience of music lovers.
 208

19.0 Tune Library and Creating Your Music Playlists

The DCart10 software provides a useful feature that you can use to enjoy the results of
your restoration work. As your restoration work progresses you will create music files in the
Restored folder and the Tune Library provides a means to find them and create playlist of
music to be enjoyed on your audio equipment.

19.1 DC Tune Library Basics

The DC Tunes software does not modify the music content or file location on your
system in any manner. When you create a music entry or delete a music entry in the library
the original music file on your system is not changed. The software in the library only points
to the file location in your computer system.
The structure of your music folders and the method that the software uses to loads in
the information must be in sync with each other. This loading in the file information requires
that you use a specific way to set up the titles in the library layout. The help file for DC Tunes
Library describes a method of loading in the music to the library when the files contain Tags
that provide needed information regarding artist, title, and more. For the Music that you have
restored using the methods in this Handbook, this information about song title and artist will
be provided by the file structure, not by Tags on the music file. The loading in the music using
the file system described in this Handbook will be used instead of the Tags feature. If inclined,
you can try the Tag method.
The Tune library’s preferences window shows the method used by the software to bring
in the music for the library when not using Tags. Refer to Figure 19-1:
 209

Figure 19-1 Tune Library Options

The box to use tags is not checked so that the path used is shown. The order of the
loading in of the file information can be modified by clicking the feature (Genre, Album,
Artist, and Title) and then selecting the move up button.
The library also provides a useful way to locate and use music that you may have from
other sources for example music CD’s. This feature for CD’s will not be covered in the
Handbook.
You can change the column names that show on your Library Window; however, these
default names will always appear within the software commands.
 210

19.2 Music Folder Layout on your System

The source of the music for the library will come from the Restored Music Files on
your computer system. If you use the methods described in Chapter 5 (Music File
Organization), then the DC Tune Library will pull in the file’s information correctly after
some modifications to the preferences section.
Under the Preferences Tab is the library path that shows, where, the actual file
information for the library is located. The DCart10 help files describe these files and it is
possible to delete some or all if you want to start over with a new library set-up.
If you have two different Diamond Cut Software products (DCart10 and DC Forensics)
you can use a common library path for each program so that the libraries are in sync. The
ability to find a specific song with this database is very useful.

19.3 Library Headings

The library software has four columns that have default names labeled: Genre, Artist,
Album, and Title. When the library is used addition column information is shown to the right
of these names providing more information about the files.
The column names used are shown below and have been changed from the default
names using the edit command in the Tune Library Preferences section. Figure 19-2 shows
the changes:
 211

Figure 19-2 Library Column New Titles

When this command is performed, the Tune Page will look like this:

Figure 19-3 Library Columns

This figure shows only the beginning of the library columns as they continue further to
the right. Now that the library columns are labeled, we can import the music. The importing
of the music will be different for LP Albums versus individual records.
 212

19.4 Importing the Music File Information into the Library

The first step will be to set the library preferences to the correct values unique for each
of the two layouts.
When the music file information is imported, the various details of Title, Artist, etc.
are entered in a specific manner by the Tune software. When you have imported a folder,
review the results in the library to understand the details of how the information came from
the folder headings with the specific and Column Title preference settings.

19.4.1 Individual Record Setting

Figure 19-4 Preferences for Individual Records

 213

Now, we will select the import folder command as shown in figure 20-5:

Figure 19-5 Importing Folder

After the command is opened the Tune Library will contain all the files in the 78
Electric Folders (Al Jolson for this example). If you have other individual records that are
Edison Diamond Disc records or 78 Acoustic, you can perform a similar operation.

19.4.2 LP Records

The preference settings are changed to the values shown in figure 19-6:
 214

Figure 19-6 Preferences for LP Albums

Use the move up command to change the layout. Now we will select the import folder
command as shown in figure 19-7:
 215

Figure 19-7 Importing LP Folders

The results of these two operations are shown in Figure 19-8:

Figure 19-8 Tune Library Results

The results for this Tune Library example can be modified for your own titles and
customized the way that you want.
 216

Reference Section
The reference section contains addition audio information for your restoration work.
The main sections of the handbook contain information that would be used often and the
reference section contains additional information that provides technical insight into audio
engineering.
As you venture into this wonderful hobby of music restoration you have many technical
sources available to add to their knowledge. This section is a beginning step into this learning
journey.
 217

20.0 History of Recorded Sound

Learning about the history of recorded music aids your restoration of the songs from
your records. The development of recording sound changed from using cylinder shaped
records and a simple recording method to the later vinyl long playing records that contain
multiple mixes of instruments and singers to create a two-channel stereo record. It is useful
for the restoration process that you understand how the music was placed into the record
grooves for the specific records you have. This chapter contains a brief over-view of this
fascinating subject.

20.1 Beginning of Recorded Sound

The first records made for sound recording used mechanical methods for both
recording and the playback of sound. The original phonograph patented by Thomas Edison
in 1878 used a cylinder made of soft tinfoil to hold the music vibrations. As the music was
being recorded the sound waves moved a recording needle that indented the soft metal in
response to the acoustic air vibrations. When the recording system was used in reverse the
music could be heard from the indentions in the metal moving the same needle. These first
machines were powered by turning a handle that rotated the record for recording and
playback.
Edison specified several different motions in the recording needle to hold the sound.
One motion was vertical, and the other was horizontal with the vertical motion known as hill
and dale and the horizontal as lateral cut.
Edison did not continue the development of the early phonograph any further at this
time as his full-time attention was directed to the development of the electric lighting system.
Other inventors picked up the idea of recording sound from Edison’s early tin-foil
phonograph and they started down two separate development paths; namely the cylinder
record and the flat record during the 1880’s. One of the early pioneers in the vertical motion
method came from a laboratory set up by the new Bell Telephone Company in Washington
DC. The lab took the tin-foil design and made many improvements to the early Edison work
with the goal of working with Edison in a future joint venture. As this early sound laboratory
grew it became the Western Electric Laboratory. Another early pioneer was Emile Berliner
who developed a flat rubber type material for the record that in turn used a horizontal motion
for holding the sound. Berliner also developed a mass production method for making his
records from a master recording which was a catalyst for developing a new Music Business
 218

for records due to this new ability to make a master recording and then duplicate the recording
in the form of flat records.
After the electric lighting industry had been created by Edison, he returned to the
development of the phonograph after learning about the work done by the Bell Telephone
Lab in Washington DC. As was his style Edison rejected working with the Telephone Lab
and went on to develop a new and improved phonograph using a Cylinder Record made of a
soft soap and wax material. The National Phonograph Company was established by Edison
to sell his improved phonograph along with various franchises to sell the new records and
machines around the world.
The National Phonograph Company initially marketed the technology as a business
machine for speech dictation. Indeed, the early sales force for the Washington District of
Columbia area was very successful selling these dictation machines to the Federal
Government. This Columbia sales franchise (named because of the District of Columbia)
stayed as an independent company during later efforts by Edison to regain sole control over
his phonograph business in the late 1880’s and would become the independent Columbia
Record Company.
Some of the early salesmen for these cylinder records found that the public enjoyed
hearing songs on the records far more than using them for office dictation. This new use for
the cylinder records (music) along with Berliners’ method for making multiple record copies
was the needed spark to the creation of a brand-new major business. As the Music Industry
was beginning, the various company lawyers and new patents caused numerous court battles.
A winner in one of the court battels took the name the Victor Phonograph Company to
celebrate their success (victor) in court. The winner (Victor) had based his new company on
the Berliner flat record patents, his own patents on a tonearm, spring motor, and his
manufacturing methods for making copies of the master recording. The speed of these records
was set to 78 RPM as this was a compromise by Berliner for music time versus fidelity of the
music. A rejected marketing picture submitted to the Edison Company of a dog listening to a
phonograph playing the dogs owner’s voice was later used by the Victor Company as a
Trademark for their phonograph machine (the artist of the picture painted over the original
Edison Phonograph with a Victor Phonograph and then submitted the picture to the Victor
Company).
The cylinder records continued to use a vertical motion for the sound waves while the
flat records used a horizontal motion to capture the sound.
These early manufactures created the record business and this new music industry. The
early pioneers in the recording industry would have to wait for the invention of the vacuum
tube, microphone, and other devices for electronic recording and playback. Therefore, the
 219

major developments for music technology were mechanical means for many years. The
general term for this time of using mechanical methods for recording and playback of music
records is referred to as the Acoustical Recording Era, since the power for the actual
generation of the record groove was provided solely by the acoustic energy from the vocalists
and their music instruments.

20.2 Western Electric Laboratory Introduces Electric Recording

During the 1920’s, new electrical inventions, were used by a team of engineers, led by
Maxfield and Harrison from the Western Electric Laboratory, to develop an electric recording
system for records. The team took the existing mechanical process used for recording of
records and added practical upgrades using the new electronic devices. This development was
finished in 1925 and the Columbia and Victor Companies used this method to record music.
The other record companies adopted similar electrical methods after 1925. In addition to the
electrical recording methods the Laboratory developed several important acoustic principals
used by the record industry for playback of music. While the recording of music was based
on electrical devices in the new Western Electric process the playback was still a mechanical
(although improved) process. The widespread use of electricity in homes had not arrived
along with reasonable priced electrical playback equipment. The introduction of electrical
playback equipment for consumers would occur later in the end of the 1920’s and into the
1930’s. The first Electrical Recorded Records were designed to be played back using the new
mechanical playback phonograph called the Orthopedic type.

20.3 History of the Edison Record Company

Early in the production of the Cylinder records the Edison Company designed a product
that could be played back often with little wear on the records. The type of stylus material
used by Edison’s phonographs to play back the music was much harder than the records so
that the needle maintained its desired shape as the records were played back. The first material
used for a Stylus was glass which was then followed by sapphire and then diamond. When
the flat records were developed Edison used a diamond stylus and hence the name Edison
Diamond Disc Records came about. A diamond stylus was also used for the 4-minute blue
Amberola cylinders. The records that Edison made started off as a soft wax and soap type of
material and then evolved to more durable materials. Edison was one of first companies to
develop hard plastic materials for the record surface and these new plastic materials were
 220

used in the new Diamond Disc and 4-Minute Blue Cylinder Records. The new Diamond Disc
Record was a flat record and allowed Edison to sell a song on both a Flat and Cylinder record
(Edison used the Diamond Disc Recording to create the same song on a Cylinder Record).
Edison’s business model was to sell a durable record with a precision, quality phonograph.
Indeed the 4-minute cylinder records played today on a restored Edison player sound very
good and are often over 100 years old! It seems that the Edison record company felt that the
customers would buy more records to increase the number of songs they had rather than
replace the records as that wore out.
The Edison Company selected the songs to be recorded in a unique manner. For the
Edison Company the main reason to issue a song was determined with how well the recording
captured the singers and instruments music; not necessarily if the music was what would sell
or be popular with the people at the time. For Edison it was important that the music sounded
as close to the original recording as possible. Since the recording methods were Acoustical
for many years the Edison songs did not include instruments that recorded poorly. The result
was that while a record may have sounded good, the sales could be poor if the public didn’t
enjoy the song.
Thomas Edison himself would often decide if a record would be produced after the
initial studio recording using his own set of standards up to the early 1920’s. While his
judgement may have been accurate about how well the recording was made using the
acoustical mechanical means, his sense on what the public wanted was not well developed.
Except for some very late records in the 1920’s Edison used a vertical motion to record the
music sound. Edison’s original record patent described both a vertical and horizontal motion
that could be used for the sound recording; however, he believed that a vertical motion
produced a better recording. The Edison records were designed to be played back at 80 RPM
for the flat Diamond Disc records in contrast to the 78 RPM speed used by the other
companies. The Diamond Disc records were a quarter inch thick and weighed one pound in
sharp contrast to the thinner and lighter 78 RPM records.
An interesting tidbit of history is that Edison was very hard of hearing for most of his
adult life and would listen at times to the record music by biting the side of the record player
and having the sound travel to his inner ear through his jaw. Isn’t it strange that the inventor
of the Phonograph was almost deaf!
Another unique feature of the Edison records that helped to reduce wear was that the
arm that carried the reproducer was driven across the record by a separate leadscrew
assembly. In this manner the groove walls of the record did not provide any force to move
the arm and therefore the groove walls maintained their shape with repeatable record playing.
 221

The Edison records were not compatible with the other record producers which forced
the consumers to purchase a specific phonograph to play back the Edison records. The result
was that the sales of the Edison Diamond Disc records were not as good as the other record
companies and as time went on, they continued to decline. As the 1920s’ were coming to an
end the Edison Company introduced a new record that used a lateral method of recording and
were thinner than the Diamond Disc records. These records were designed to compete with
the traditional records made by the other companies and used electrical recording and low
frequency modification (turnover). However, these records did not sell well at all. The
inventor of the Phonograph, Thomas Edison, stopped the manufacture of Records just before
the Stock Market Crash in October of 1929.

20.4 Victor Company and Others Business Models

Most of the companies that were competitors to Edison used a different business model
for their products. Their records were flat and made of a hard material that had as its primary
component ground-up rock held in place with a binder of shellac. The needles that were used
in the reproducer were made from soft steel and had an initial sharp point that was not
designed to fit the record groove. The reason for the sharp point was that the needle or stylus,
being softer than the record material, would quickly wear down to the correct shape of the
specific groove for that record. The initial wearing in was to happen within a couple of
revolutions of the record. Since that needle was now shaped to the specific groove for that
record the rest of the song on the record would sound good. As you would expect you should
replace the needle after each new record that was played. Many steel needles were sold!
The fact that the records were hard enough to wear down steel needles meant that metal
particles are often present in the record grooves from the initial wear in period. In addition,
while the record was made of a hard material, groove wear would still occur with each
playback as it shaped the steel stylus.
The motion that drove the tone arm containing the reproducer was driven by a force
provided by the groove walls of the record. The Victor Company and other record companies
derived the needed force to move the tonearm directly from the same grooves that contained
the recorded sound. This caused additional wear to the groove walls as the records were
played.
The business model for the Victor company and others was to provide a low-cost
phonograph to customers with a resulting trade off that the records would not last as long as
 222

the Edison (Competition) records. By having the records wear out and be low in cost the
customers would buy more copies of their favorite records to replace the ones that worn out.
The Victor Company was aware of the public’s music tastes and produced many
records that followed the current trends in songs. The records were marketed and sold to
satisfy the people’s desires in contrast to Edison’s approach to produce a recording that
reproduced the instruments as good as possible. Therefore the 78 RPM records often had the
first jazz music and would introduce many future singers and players in the field of recorded
music. The Edison records, in contrast, didn’t have many of the newer types of music and as
many new singing artists until the early 1920’s (Edison’s direct involvement with song
selection ended in the early 1920’s).
In the late 1920’s the owner of the Victor Record Company sold his company to the
RCA Radio Company. The new RCA Victor Company would continue for many years to
produce music long after Edison’s Company had stopped making records.
The early Columbia Phonograph company, started by selling cylinders for office
dictation, also continued for many years to produce music.

20.5 New Field of Audio Engineering Created

The pioneering work by the Western Electric Laboratory in the science of electric
recording started a new engineering effort into audio recording and playback. This Era of
Audio Engineering after 1925 benefited from inventions and new low-cost electrical devices
that were rapidly developed for the Radio, Phonograph, and Motion Picture businesses. A
careful reading of this time in the audio technical journals reveals the early work that was
started towards Stereo Recording, Vinyl Low Noise Records, FM radio, and true high-fidelity
performance. Much of the audio engineering work was delayed till after World War Two
ended. When the War was over the new electrical technology from the War, coupled with the
high demand for consumer goods, led to an explosion in the quest for High Fidelity. For much
more information the technical magazines (Audio Engineering) and the numerous College
Textbooks from the 1940’s and 50’s should be consulted. Many well-known audio companies
were started in this time to satisfy the demand for high fidelity audio such as Heathkit, Scott,
Marantz, Fisher, McIntosh, and many more.
The new Tape-Recording technology was used to produce the Master Recordings in
the studio; along with continuing refinement of phonograph records led to the Stereo Long
Playing Records we have today. The next step to the Digital Storage of the music was not so
 223

much as an improvement in the Recording of the Music as it was an improvement in the

manufacture and distribution of music.

20.6 Can Edison Records be played on Victor Phonographs?

Both the Edison and Victor Music companies felt that their records used the correct
technology for recording sound. As you would expect their records were not compatible with
each other’s record players. Unfortunately, an Edison Diamond Disc record would be
seriously damaged when a steel needle with the wrong stylus shape was used to play the
record as would happen when using the Victor phonographs to play an Edison Flat record.
The used record stores will often have Edison Diamond Disc records which appear to
be O.K. but when played back have a scratchy sound for the first moments of the song. These
few seconds was the amount of time for a person to realize that a steel needle was the wrong
stylus to be used for playback!
With the current low mass turntable arms and diamond stylus used in todays’ turntables
the Edison Flat records are not damaged when now played back, but they will sound rather
thin and low in volume before the correct vertical stylus motion is selected.
 224

21.0 Recording Details for Music

21.1 Physics of Sound

Some physics principals about acoustic waves will be used in this section to understand
the electrical recording process. Many texts and resources are available to understand the
physical concepts to a deeper level. The first step is the general Acoustic wave equation that
describes the motion of the displacement particles from a vibrating source of sound.
The equation for the displacement or amplitude of the particles in an acoustic wave at
any instant and point in space for a wave traveling in the positive x direction is given by the
equation:
𝑉𝑚 𝑥
𝜉 = ( ) sin 𝜔 (𝑡 − ) 𝑒𝑞 (21 − 1)
𝜔 𝑐

Where ξ = the displacement of a particle from its normal position

Vm = the maximum particle velocity (a constant value)
ω = 2πf, with 𝑓 being the vibration frequency
c = velocity of sound
t = time
𝑥 = distance from source
The sine function is for a pure tone.

It is assumed that the maximum particle velocity will be constant and independent of
frequency. The velocity of sound in air is about 1,130 ft. per sec. for room temperature and at
sea level. If the equation is used to describe a particle at the source itself, then x/c is equal to
zero.
The relationship between frequency and displacement of such a vibrating source, for
the same applied force and the same power, is an inverse one (Vm/ω). What this means is that
as the frequency of a vibrating source goes up the amount of displacement required to generate
the same amount of sound power decreases proportionately and vice versa the lower the
frequency the greater the displacement for the same power. This inverse relationship is central
to understanding the need for amplitude modification to the recorded frequencies for records.
The velocity term will now be used to describe the change in the displacement waveform
versus time. This velocity term does not refer to the speed that the sound travels through the
air (c).
 225

21.2 Constant-Velocity and Constant-amplitude

The velocity of a sine wave has a maximum value as the waveform is crossing the zero
axes. The average velocity is defined as the displacement per time over a cycle of the
waveform. For a constant sound power of force, the maximum displacement will decrease for
an increase in frequency as an inverse relationship and the velocity of the waveform will
remain constant for the frequency change. The phrase constant-velocity for constant signal
power is shown using figure 21-1:

Displacement vs Time
1.00000

0.80000

0.60000

0.40000
Displacement (inchs)

0.20000
50 Hz P-P 2.0
0.00000
100 Hz P-P 2.0
0.000 0.005 0.010 0.015 0.020
-0.20000 100 Hz P-P 1.0

-0.40000

-0.60000

-0.80000

-1.00000
Time (Sec.)

Figure 21-1

The first waveform shows a 50 Hz sinewave with a displacement of plus and minus 1.0
inch, next are waveforms at 100 Hz with displacements of 2.0 and 1.0 inch.
As we increase the frequency of the 50 Hz sinewave, while keeping the power of the
waveform the same, there are two options that can show a change in frequency as either the
100 Hz P-P (Peak to Peak) 2.0 or 100 Hz P-P 1.00 waveforms. The 100 Hz P-P 2.0 waveform
keeps the displacement the same but now the velocity of the signal as it crosses the zero axis
 226

is twice as steep (slope is 2X), while the 100 Hz P-P 1.0 waveform has the amplitude reduced
by a factor of two and the velocity as it crosses the zero axis is the same as the 50 Hz
waveform. The velocity must stay the same for constant power so, the change from the 50 Hz
2.0 P-P to 100 Hz 1.0 P-P waveforms demonstrates how the amplitude changes for constant-
velocity factor while the same signal power remains.
These terms apply to the tip of the recording or reproducing stylus as it traces the record
groove. In constant-amplitude recording, the stylus tip at a constant signal power moves a
fixed distance each side of its center or rest position for any frequency. The amplitude of the
swing of the stylus tip is constant for all the frequencies in this region for the same amount of
sound intensity. For example, if the sound intensity drops, then the amplitude of the swing of
the stylus will drop. However, if the sound intensity stays the same as the frequency changes,
then the amplitude swing value of the stylus will stay the same.
To demonstrate constant-amplitude let’s look at figure 21-1 and now start with the
signal at 100 Hz and amplitude 2.0 P-P. If we keep the swing of the stylus tip the same as we
decrease the frequency to 50 Hz and amplitude 2.0 P-P, then that is an example of constant-
amplitude. What is important to understand is that the power of the signal at 50 Hz is now
less than the power at 100 Hz to keep the swing or displacement the same. This is because
the velocity of the 50 Hz waveform is half that of the 100 Hz waveform and the velocity is
related to the power. The use of constant-amplitude recording will require a modification to
the signal as it is recorded that will be dependent on its frequency.
In constant-velocity recording, the maximum velocity of the stylus tip at a given signal
level remains constant for any frequency as was shown. The use of constant-velocity will not
require a modification to the signal as it is recorded that is dependent on its frequency.

21.3 Constant-Acceleration (Pre-Emphasis)

Constant-acceleration is another term that helps to describe a type of modification to

the frequency response of the music that is useful for surface noise reduction. Today this term
is called Pre-Emphasis instead of constant-acceleration. Let’s look at figure 21-1 and examine
the difference between the 50 Hz P-P 2.0 waveform and the 100 Hz P-P 2.0 waveform. The
velocity is changing, and the energy is increasing from the 50 Hz to the 100 Hz waveform. If
we cause this to happen above a certain frequency by increasing the energy in the signal at an
increasing constant rate then since the velocity is steadily increasing, we can call the result
constant-acceleration.
 227

The reason that this method can help the recording of music is related to the surface
noise on the record. The surface noise on the record is mainly random noise with uniform
noise frequency. As music frequencies increase, the amplitude of these frequencies drops to
a point where the random noise and the music are similar in amplitude. Now the music
frequencies are hard to hear above the surface noise. By steadily increasing the amplitude of
the recorded frequencies (above a certain value) during the production of the master record
the music rises above the noise. When the record is played back the increase in amplitude will
be reduced to the correct level, but during this reduction, the surface noise will also be reduced
by a corresponding amount. The net result is that the range of the upper frequencies that can
be heard from the music has been extended upward by reducing the surface noise that was
limiting the hearing of these music sounds.
The implementation of this method specifies the time constant of the circuits rather
than a specific frequency for this change in music power to occur. However, the reverse
specification can be either in time constant values or in frequency. The term roll-off refers to
a frequency where the correction curve starts and the amount from a reference point.
This constant-acceleration concept to extend the upper range of recorded music was
discussed in the original work by the Western Electric Labs in 1925 but was not used on a
regular basis until many years later at which time the term Pre-Emphasis replaced the original
term. My opinion on why the term Pre-Emphasis was used for records is that prior to the wide
use of this concept for records the FM radio transmitters and receivers had developed a similar
method (boosting the high frequencies during transmission followed by reduction at the
receiver) and they referred to their concept as Pre-Emphasis. Hence the record companies
used the same term as it was familiar with the public.

21.4 Early Cutting Grooves on Discs

The early sound recording engineers did not have many tools or reference information
available to them as they created the first commercial records. Many trial and error
experiments occurred using mechanical equipment for recording and playback. All Acoustic
recording used a horn to both capture the sound and match as best as possible the audio
characteristics of the recording groove cutting device.
The Acoustic Era recording engineers provided damping in the recording mechanics
to produce a constant-velocity recording for the music power from 400 Hz to 3000 Hz. Below
400 Hz the motion was constant amplitude. The early records using the acoustic process had
 228

a turnover around 400 Hz due to the technology, not the desire. This constant-velocity
recording was true for both the Vertical and Lateral motion for the record grooves.

21.5 Playback of Acoustic Era Records

The magnetic cartridge that is used to convert the groove motions to sound produces a
signal due to changes in magnetic flux. These signal levels are directly proportional to the
velocity of the stylus (flux changes) as it follows the groove motion. Therefore, the magnetic
cartridge is a constant-velocity device. If the input to the magnetic cartridge had a constant
amplitude as the frequency was increased, then the output versus frequency would rise at a 6
dB per octave rate. This distortion can be seen when cartridge specifications are shown
(indicating where the cartridge deviations from the desired shape). However, the recorded
music sound on a record, using constant-velocity recording, has a decreasing amplitude with
frequency and the use of a magnetic cartridge with a constant-velocity recording will provide
a resulting flat response over frequency. This is not true for the crystal type of pickups that
were occasionally used as they are sensitive to the displacement not the rate of change of
displacement.
The lower frequency amplitudes with acoustic recording did not cause a problem in
recording since they were recorded with very low amplitude due to the horn technology. This
was not true when electric recording was introduced.

21.6 Amplitude Modification with Frequency for Electric Recording

The previous sections provide the groundwork for the understanding that there will be
real problems in directly recording music into the record grooves without some type of
amplitude modifications to the new low music frequencies present due to the use of
microphones and electrical amplifiers.
For constant acoustic power the amplitude or the swing of the cutter increases as the
frequency is decreased and a point is reached, as the music frequencies decrease, where there
is a chance that one groove would cut into the next. The space between the grooves cannot be
changed without decreasing the playing time of the record and since the record companies
had an existing product being sold (acoustic recorded records) with a certain playing time
something had to be done to make the new electric process work with the existing record
groove spacing. The previous acoustic recording method did not reproduce low frequencies
 229

very well so that the increasing amplitude motion with lower frequencies had not been a
problem with the cutter touching the adjacent groove during the Acoustical Recording Era.
The solution that was developed by the Western Electric Laboratory (Maxfield and
Harrison Team) was to modify the bass frequency content of the recording signal going to the
cutting mechanics. Progressive bass attenuation was introduced below a chosen value called
the turnover frequency resulting in the amplitude of the cutter’s swing for all signals below
this frequency remaining the same for the same power of the incoming signal. This
relationship is called constant-amplitude (for constant power in the frequencies). In practice
this technique is implemented with modifications made to the equipment’s frequency
response using a 6dB per octave attenuation rate (or 20 dB per decade) starting at the turnover
frequency. As the frequencies decrease from the turnover value the amplitude of the signal to
the cutter is reduced further and further by the same amount that the amplitude would have
been rising for constant input power. By using this modification to the lower frequencies, the
same record spacing/music time was maintained for the new records as for the previously
manufactured records using the previous acoustic recording.
For frequencies above the turnover value the grooves are cut with a constant-velocity
characteristic which means that no intentional changes are made to the music since this is the
natural or normal way that the cutter works, and that sound wave behaves.
The other method constant-acceleration, while discussed in the original research paper,
was not implemented at this time (1925).
You can divide the frequency range of both the recording and playback process into
sections that have either constant-amplitude or constant-velocity characteristics.
When the RIAA curve became standard for records the frequency range of both the
recording and playback used three distinct areas. The lowest range was for constant-
amplitude, the middle for constant-velocity and the upper for Pre-Emphasis or constant-
acceleration.

21.7 Notes on Filter Response

When filters are described, a typical description will state the amount that the filter
changes per frequency interval (slope). For the typical electrical filters used in audio work a
value of 6 dB per octave (factor of 2) or 20 dB per decade (factor of 10) is often stated. These
filters include a second value to say where the change in amplitude starts or where the slope
changes from 0 to plus or minus 6 dB per octave. The value where the slope change occurs
will be given as a frequency but this change in slope does not happen exactly at this value. In
 230

an actual filter the slope of the filter’s amplitude versus frequency change starts before the
quoted frequency value then changes plus or minus 3 dB at the quoted frequency; and
continues to change slope until it equals the 6 dB value. The change is not a sudden change
rather it’s gradual. In the graphs that show the filter’s response (Paragraphic EQ) you will see
this gradual change occurring. Remember that the frequency of the filter is quoted at the 3 dB
change point for the common 6dB slope filters.

21.8 Early Electric Recording & Playback Conditions

The work by Maxfield and Harrison had several goals for accurate and faithful
performance across the range of audio hearing. However, as early pioneering work had to
deal with the state of the art at that time, the actual performance fell short of their many goals.
Since the technology prior to their work was a total mechanical system driven by sound power
their first product was much better than anything before. The fact that a sensitive microphone
was used to record the sound meant that the musicians could perform in a studio under the
normal conditions used musicians for the location of instruments, rather than in an artificial
setup for the acoustical recording in which performers shouted directly into the recording
horn and pianos were placed up on boxes to balance the recording sound. This new technology
allowed drums and bass instruments that had previously been difficult to record to now be
used in song arrangements. These new electric recorded songs caused a major revolution in
the music world.
At the time of the development of this new Western Electric Recording System the
average record buyer did not have access to consistent, low cost, electrical power in their
home. In addition, any electrical equipment that could be used for playback of the new records
had yet to be invented and mass produced. While the record manufactures could afford to
spend money for the latest technology for recording the music the consumers of the records
couldn’t. Thus, the Western Electric company introduced an improved mechanical playback
of the records that used no amplification by electronics. This new mechanical system
reproduced a larger range of the recorded frequencies now present on the new electric records.
They improved the recording of the records but did not change the playback from a
mechanical system for the early electric recorded records. When you find records from this
early electric era, the record label will often have the words electric recording written with
lightning bolts or other visual symbols of this new technology. The very early Victor records
will often have the words Orthophonic Recording and VE on the label which means that you
would use a new mechanical phonograph.
 231

The initial recording system developed by Western Electric had a response from 50-
5000 Hz with a natural turnover at 200 Hz due to the mechanical performance of the cutting
system for the master. The change from constant-velocity to constant amplitude around 200
Hz was due to the best that the technology could do at that time (1925). The potential
modification to the high frequencies for surface noise reduction was not implemented since
the upper frequency limit possible was still below surface noise. The first electrical recordings
couldn’t take advantage of all the design concepts due to the manufacturing technology
available at the time.
The mechanical devices that cut the grooves in the master record are carefully designed
such that the resonances in their structure do not occur within the audio frequencies by having
the Q or Quality Factor of the mechanical cutter set to a value around 0.1 which is an under-
damped system. What this does is to make the velocity of the cutter (the sharp edge that makes
the grooves) the same or constant for the same applied power over the widest audio frequency
range possible. You do not want any mechanical resonances to amplify or change the sound
of the music. The original work by the Western Electric team (Maxfield & Harrison)
described how they were able to achieve this result using mechanical damping (rubber line)
and other methods.

21.9 Later Recordings

The initial work by the Western Electric company changed the record industry in a
major way. The continuing rapid development of electrical devices expanded the audio range
that was recorded and the record companies were able to introduce new styles of music since
the microphone allowed a wide range of instruments to be used in songs. For singers the
microphone allowed a crooning style that became very popular. However, a significant
opportunity was lost at this early electric recording moment to come up with a set of standard
settings for the turnover value and potential pre-emphasis values across the record industry.
Due to the human nature of the recording engineers, each record company felt that their
setting for the recording of the music for turnover and use of pre-emphasis were the correct
values to use. The specific values were at times viewed as trade secrets and not widely
described.
The recording engineers at the record companies used unique settings to make the
record sound good for the customers by understanding that few phonographs at the time (20’s
& 30’s) were high performing audio devices. Some records had extra high frequency and
extra low frequency boost added so that the music would sound good on a typical customer’s
 232

phonograph when played back. The same record that would have sounded good on a 1938
phonograph may not sound correct on a current high performance sound system that has a flat
response from 20 to 20,000 Hz.
Much later when the new 45 RPM record and LP record were developed, the various
record companies realized that a standard playback EQ (Equalization) response was needed;
which established a uniform standard (RIAA) for the turnover and pre-emphasis values.
During the time 78 RPM records were made no industry wide set of recording standards
existed. There were even differences between the European and American manufactures.

21.10 Edison Electrical Recording

Even though the Edison Company had pioneered the recording of sound they did not
adapt to the new electrical methods until much later after the initial work by Western Electric.
The various engineers within the Edison Company tried to bring in this new technology but
the founder of the industry, Thomas Edison, did not want to change the business model from
continuous improvement in the acoustic recording method to something new. The records
made in 20’s by the Edison company are very good recordings, although limited by the
Acoustic Recording process in the types of instruments that were used. By 1927 the Edison
Company developed their own methods for electrical recording with equipment from the
General Electric Company (GE was established to manufacture equipment from the use of
Edison Patents). Much of the specific information about Edison’s Electric Recording work
was not published and detailed information is not available. What is known is that the
Diamond Disc Records had a much fuller sound around 1927.
Research by the Diamond Cut Productions Company found that a turnover value of
500 Hz had been incorporated into the Edison Phonograph and Radio that was made for a
brief time in the late 1920’s for playback of the new Needle Type Edison lateral records. At
this point the Edison Company was attempting to refresh their product lines to the current
state of the art, but they were too late and in October of 1929 (just before the stock market
crash) Thomas Alva Edison, the inventor of the Phonograph, shut down the manufacture of
all records.
In order to playback records correctly, knowledge of the recorded electrical properties
will be needed so that the modifications to the recorded sound can be removed. If the exact
opposite to the modifications made during the recording of the master record are used, then
the original desired sound will be heard.
 233

22.0 Selection of Cartridge and Stylus for Transcription

22.1 Introduction

The selection of the cartridge and stylus for taking the sound from the record groove
and converting it to an electrical signal is critical to the success of music restoration for
several reasons:

1. The transcription process cannot damage the sound grooves since the record has
value as a piece of history.
2. The process must extract the maximum signal and the minimum noise from the
record groove. Noise is defined as both surface noise and distortion to the original
sound.
3. The conversion of sound to mechanical motion used different methods by the music
companies which in turn require unique approaches for transcription.

22.2 Cartridge Selection Acoustic and Electric Recording

The Acoustic recording method for lateral and vertical groove motion was a constant-
velocity type over the main range of recorded music. The low frequency end of the music had
reduced energy due to the use of the mechanical recording equipment with provided a natural
turnover to constant amplitude in the general range of 400 Hz.
The Electric recording method uses a combination of constant-velocity and constant-
amplitude with the turnover frequency controlled to a fixed value versus the acoustic
technology.
For constant-velocity recording, the use of a magnetic cartridge will provide the best
method for playback of the groove for either vertical or lateral motion. The magnetic cartridge
has a constant-velocity response and when used with the same recording method yields the
correct flat frequency response. Acoustic recording and Electrical recording can use the same
magnetic cartridge for playback from the record.
There are many magnetic cartridge manufacturers available, and the choice of cartridge
should provide a method to change the stylus for the different types of records you want to
transcribe.
 234

22.3 Errors in Playback from Stylus Shape for Lateral and Vertical

Audio engineers have created a very large amount of information regarding potential
sound problems for the complete recording path from the original recording to the playback
of the music for the listener. A review of this audio information provides important insight
into the best method for playback of Diamond Disc records and 78 RPM lateral recordings
using the current equipment. The first subject is how does the shape and size of the stylus
affect the original recording? One method considers that the recorded sound wave in the
record groove has a finite wavelength, defined as the distance from the beginning of the sound
wave to the end of a cycle. While this definition is correct strictly for sine waves not sound
waves; it can be used to estimate the relationship between the stylus size and the maximum
recorded frequency in the following manner.
As the radius of the record changes, the velocity of the stylus tip following the sound
groove will also change for a constant record speed. For the worst-case situation, consider the
velocity at the inner radius and the highest sound recorded. If the length of the wavelength of
the sound in the record groove is the same or close to the size of the stylus in contact with the
groove, then severe distortion will occur since the shape of the wavelength cannot be
accurately traced by the stylus. Although this concept is a starting point, an approach that
considers measuring the distortion is needed since the sound is degraded before its complete
loss.
The general term that is used by audio engineers to describe the behavior of the stylus
as it follows (or traces) the music groove is Tracing Distortion. This playback problem is
fundamentally caused by cutting the original music groove with a chisel shape (wide for the
groove width and then narrow on the edges) while a spherical shape is used to playback or
trace the music. The general tracing problem was studied in a landmark technical paper in the
Journal of the Acoustical Society of America written by W. D. Lewis and F.V. Hunt in the
late 1930’s. In this technical paper the exact motion of the stylus was calculated as it followed
the groove so that a clear understanding of the distortion caused by the stylus shape and
recording process could be mathematically described. After this article was written, several
researchers used the results to understand the limits in both vertical and horizontal recording.
The next section uses an approach by Frayne and Wolfe that built on the original work by
Lewis and Hunt.

22.3.1 Pierce and Hunt Analysis

 235

The following information has been based on the book “Sound Recording” by Frayne
and Wolfe contained in section 13-7 in their book. This reference should be consulted for
greater details if needed for tracing error. Figure 22-1 shows a spherical stylus as it travels
along a hill and dale record groove from the book. Figure 22-1 is used to describe the lateral
motion if the figure is turned 90 degrees.

Figure 22-1 Stylus as it Travels the Groove

The dashed curve shows the shape seen by the stylus and sent to the cartridge and the
solid curve shows the original sound curve cut into the material to form a groove. Notice that
the original cosine wave shape traced is distorted because the contact location changes from
the bottom of the stylus (x = 0) to a side (x) and then back to the bottom. The traced curve
has been distorted from the original and the general term is called tracing distortion or error.
The reference book continues to define the various variables and develops a method to
describe the traced curve as a series of complex equations. The authors of the book cite the
original work on this subject developed by Pierce and Hunt in the 1930’s. For the figure
shown in 22-1, the variables as defined in the book, section 13-7, are:

𝑟 = 𝑟𝑎𝑑𝑖𝑢𝑠 𝑜𝑓 𝑠𝑡𝑦𝑙𝑢𝑠
𝜆 = 𝑤𝑎𝑣𝑒𝑙𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑟𝑒𝑐𝑜𝑟𝑑𝑒𝑑 𝑠𝑜𝑢𝑛𝑑
 236

𝜃 = 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑤𝑖𝑡ℎ 𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑡𝑜 𝑝𝑒𝑟𝑝𝑒𝑛𝑑𝑖𝑐𝑢𝑙𝑎𝑟 𝑜𝑓 𝑠𝑡𝑦𝑙𝑢𝑠

𝜂 𝑎𝑛𝑑 𝜉 𝑎𝑟𝑒 𝑡ℎ𝑒 𝑠𝑡𝑦𝑙𝑢𝑠 𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠
𝑥 𝑎𝑛𝑑 𝑦 𝑎𝑟𝑒 𝑡ℎ𝑒 𝑟𝑒𝑐𝑜𝑟𝑑𝑒𝑑 𝑐𝑜𝑠𝑖𝑛𝑒 𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠
𝑎 = 𝑎𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒 𝑜𝑓 𝑐𝑜𝑠𝑖𝑛𝑒 𝑎𝑛𝑑 𝑘 = 𝑎 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

2 𝜋𝑥
The recorded sound curve is: 𝑦 = 𝑎 cos = 𝑎 cos 𝑘𝑥 𝑒𝑞 (22 − 1)
𝜆

𝜉 = 𝑥 + 𝑟 sin Θ eq (22 − 2)

𝑎𝑛𝑑 𝜂 = 𝑦 + 𝑟 cos 𝜃 = 𝑎 cos 𝑘𝑥 + 𝑟 cos 𝜃 𝑒𝑞 (22 − 3)

Where − tan 𝜃 = −𝑘𝑎 sin 𝑘𝑥 𝑒𝑞 (22 − 4)

𝑘𝑎𝑟 sin 𝑘𝑥
𝜉=𝑥+ 𝑒𝑞 (22 − 5)
√1 + 𝑘 2 𝑎2 𝑠𝑖𝑛2 𝑘𝑥

𝑟
𝜂 = 𝑎 cos 𝑘𝑥 + 𝑒𝑞 (22 − 6)
√1 + 𝑘 2 𝑎2 𝑠𝑖𝑛2 𝑘𝑥

These last two equations describe the motion of the stylus as it traces the groove, hence
the potential tracing error. More mathematical development occurs in the reference chapter
in solving these equations using a Fourier Series Expansion and then developing results for
the harmonic distortion in the sound caused by the playback mechanisms. An important result
in the reference book states: “It is evident that the shape of the poid (the traced curve) and
hence its percentage harmonic content does not depend on the actual dimensions of the
original cosine curve and the radius of the tracing circle, but rather its shape depends on the
relative values of certain of the dimensions. Thus, the shape of the poid can be entirely
specified by giving the values of the ratios of (𝑎⁄𝜆) and (𝑟⁄𝜆) where 𝑎 is the amplitude of
the wave being traced, 𝜆 is the wavelength on the record, and 𝑟 is the stylus-tip radius”.

This result gives us great insight into the correct stylus to use.

The next section in the reference develops a chart that shows total harmonic distortion
results for various conditions, namely modulation (lateral or horizontal), stylus radius,
 237

amplitude, frequency, and record speed. All these values can be reduced to a plot of 𝑘𝑎 vs
𝑘𝑟. Where:

𝑘𝑎 = ( 2𝜋𝑎⁄𝜆) 𝑎𝑛𝑑 𝑘𝑟 = ( 2𝜋𝑟⁄𝜆) 𝑒𝑞 (22 − 7)

The chart of Total Distortion using these values is shown in figure 22-2.

Figure 22-2 Constant Distortion Contours

Notice on the chart that there are lines for both Vertical (Dashed) and Horizontal (Solid)
motion and that the Vertical values are greater than the Horizontal for similar conditions
(cancellation of even harmonics occurs for horizontal motion).
The analysis in the book continues with general development of other relationships that
can be established with the following conclusions.
 238

The Total Harmonic Distortion Varies:

1. Directly with the recorded amplitude for vertical and as the square of recorded
amplitude for lateral records, assuming constant frequency, groove speed, and stylus
radius.
2. Directly with frequency for vertical and as the square of frequency for lateral, other
qualities remaining constant.
3. Inversely as the groove speed for vertical, inversely as the square of groove speed
for lateral, other qualities remaining constant.
4. Directly as the stylus radius for vertical, directly as the square of the stylus radius
for lateral, other qualities remaining constant.

These relationships can be written with these equations:

𝑟𝑎𝑓
𝐷𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑣𝑒𝑟𝑡𝑖𝑐𝑎𝑙 𝑟𝑒𝑐𝑜𝑟𝑑𝑠 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 × 𝑒𝑞 (22 − 8)
𝑉𝑔

𝑟 2 𝑎2 𝑓 2
𝐷𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑙𝑎𝑡𝑒𝑟𝑎𝑙 𝑟𝑒𝑐𝑜𝑟𝑑𝑠 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 × 𝑒𝑞 (22 − 9)
𝑉𝑔2
Where
(2𝜋𝑅)𝑥 𝑅𝑃𝑀 𝑖𝑛𝑐ℎ
𝑉𝑔 (𝐺𝑟𝑜𝑜𝑣𝑒 𝑆𝑝𝑒𝑒𝑑) 𝑖𝑛𝑐ℎ⁄𝑠𝑒𝑐 = ⁄𝑠𝑒𝑐 𝑒𝑞 (22 − 10)
60

The book concludes with a graph, figure 22-3, showing Total Distortion versus
Recorded frequency for a spherical stylus with a radius of 2 mils and 2 mils of constant
amplitude up to 300 Hz and constant-velocity above 300 Hz. These values represent some
lateral record conditions at the time of writing the book.
Notice on the graph the equation that calculates the amplitude change (decreases) as
the frequency rises from 300 Hz (turnover value).
 239

Figure 22-3 Distortion for Various Record Radius

Now that this information is available, calculations for harmonic distortion for Edison
Diamond Disc Records and 78 RPM records can be performed to understand the effect of
stylus size on performance. Two cases for vertical diamond disc records will be studied using
the original stylus shape by Edison and a current stylus used by DJs for LP records. The DJ
stylus was selected from experimental studies that found that a current product from the
Ortofon company performed very well with the diamond disc records and exceeded the
original Edison design for clean sounding music. The lateral records will use a 2.5 mil stylus
and a 3.5 mil stylus where the 2.5 mil was originally recommended by cartridge companies
for “typical” 78 RPM records and the 3.5 mil was from experimental studies. First the original
diamond disc stylus design.
 240

22.4 Edison Diamond Disc Vertical Modulated Records Stylus Selection

Diamond Discs records used a hill and dale method to store the sound waves in the
record groove. This method had been used by the Edison Company for the cylinder records
and continued with some changes for the new flat records. Figures 22-4 and 22-5 are from
the Edison National Park and shows the diamond stylus and the stylus in a record groove.

Figure 22-4 Diamond Disc Stylus

 241

Figure 22-5 Diamond Disc Stylus in Groove

These two drawings provide important information regarding the correct stylus to use
for the records. Since the modulation of the sound is vertical and the maximum depth is 1.74
mils (thousandth of an inch), the midpoint would be half of this value or 0.87 mil and would
correspond to no sound. The peak movement then would be ± 0.87 mil which is a very small
amount.
The radius of the diamond point as shown fits within the groove and is designed to ride
on the bottom of the groove. The music information is contained at the bottom of the groove
and the width of the groove will vary as the recording cutter travels up and down. The original
groove on the record was cut into a soft master recording using a shape described in Edison
Patent number 964,221. This patent explains a new method of recording the sound that
deviated from the previous cylinder recordings by using a smaller groove size and smaller
depth to improve the performance of the cutter so that downward and upward cutting motions
would need similar cutting energy. The previous method introduced some distortion by
having more sound energy needed to cut down than up.
 242

Figure 22-6 contains drawings from the patent that shows the reduction in the groove
size using the new method as shown in patent fig. 4, the diamond disc groove and patent fig.
3, the previous groove on the cylinders.

Figure 22-6 Patent Information

The Edison Music Company, as did the other Music companies, did not publish
technical information regarding the details of their record production. Changes occurred due
to constant improvement so that the drawings and Patent information can be used only as a
guide.
Since the Diamond Disc Phonograph used no electrical amplification until much later
at the end of the 1920’s the motion of the sound grooves had to be amplified via mechanical
methods. Within the Edison reproducer was a lever that amplified the groove motion coupled
to a heavy floating weight so that when moved by the stylus motion would in turn move a
special diaphragm to produce sound waves. The force applied to the stylus was approximately
190 Grams and could be more for different models. This heavy force moved by the groove
would then produce a significant amount of sound energy for the listener. Because of the large
stylus force, Edison used a rather large diamond point in a spherical shape to provide a
reasonable lifetime for the records, as the pressure on the record groove is related to the
applied force and the size of the contact area. The diamond point used had a nominal radius
 243

of 3.5 mils which was in full contact with the record groove. These records had a thick core
(about one quarter inch thick) with a hard plastic material on the surface that held the sound
groove. When a transcription is made today with electrical amplification the same stylus force
and shape originally used does not have to be used. What are the limitations in the
reproduction of the music from the shape and size of the stylus?
To start, how was the music groove cut in the recording blank. The recording at the
music studio was made with a soft material and a cutting tool that was driven by the acoustic
energy from the musicians. After this original recording was made, the master stamping
molds for the actual records were copies from this original recording. When the original
recording was made a Patent number 1,024,839 from Edison provides some information about
the groove shape. Figure 22-7 shows some drawings from this patent.

Figure 22-7 Recording Cutter

In the patent description, Edison refers to the shape in the patent Fig. 2 as that of the
head of a pin with a cut made into the surface. From the patent you can see that the groove
shape is like a chisel with the wide portion cutting up and down with the song as the disk
rotates. The question is what is the result of using a certain stylus shape as it rides in this
record groove cut with the recording cutter? The results of the Pierce and Hunt Analysis
Method are used to examine the performance of the original diamond disc stylus and a stylus
 244

found from experimental results, namely the Reloop DJ 0.7 OM Black from Ortofon. This
stylus from Ortofon is a current DJ style that has a 0.7 mil radius on a spherical polished
shape.

22.4.1 Worst Case Calculations for Two Stylus

The worst-case playback conditions for the Diamond Disc during a playback are found
at the inside record grooves (close to the center) and for high frequencies of short wavelength.
For the calculations, two high frequencies of 4 kHz and 8 kHz and two stylus sizes will be
used; the original Diamond Disc 3.75 (Max) and a 0.7 mil DJ (Disc Jockey) Spherical Shape
designed for “Scratching and Playing” LP records.

First Case: Inside Track (2.38 inches) and 8 kHz sound wave.

(2𝜋𝑅)𝑥 𝑅𝑃𝑀
𝑉𝑔 (𝐺𝑟𝑜𝑜𝑣𝑒 𝑆𝑝𝑒𝑒𝑑) 𝑖𝑛𝑐ℎ⁄𝑠𝑒𝑐 = = 19.94 𝑖𝑛𝑐ℎ⁄𝑠𝑒𝑐 𝑒𝑞 (22 − 10)
60

𝑓𝑜
𝑎 (𝑝𝑒𝑎𝑘 𝑎𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒) = 0.87 𝑚𝑖𝑙 𝑥 ( ⁄𝑓)= 0.04 𝑚𝑖𝑙 𝑒𝑞 (22 − 11)

Where 𝑓𝑜 = 400 𝐻𝑧 (𝑊ℎ𝑒𝑟𝑒 𝑀𝑎𝑥 𝐴𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒 𝑂𝑐𝑐𝑢𝑟𝑠) 𝑎𝑛𝑑 𝑓 = 8000 𝐻𝑧 (note

that constant-velocity recording was used so that the amplitude will be reduced as the
frequency increases per eq (22-11).

𝑉𝑔
𝜆 (𝑅𝑒𝑐𝑜𝑟𝑑) = = 0.00249 𝑖𝑛𝑐ℎ 𝑒𝑞 (22 − 12)
𝑓(𝐻𝑧)

2𝜋𝑎 2𝜋𝑟
𝐾𝑎 = 𝑎𝑛𝑑 𝐾𝑟 = 𝑒𝑞 (22 − 7)
𝜆 𝜆

For 3.75 mil (Original Diamond Disc Stylus), 𝐾𝑟 = 9.46 𝑎𝑛𝑑 𝐾𝑎 = 0.10 which gives
from figure 22-2 a total distortion value ≈ 50 %.

For 0.7 mil (DJ LP Stylus), 𝐾𝑟 = 1.77 𝑎𝑛𝑑 𝐾𝑎 = 0.10 which gives from figure 22-2
a total distortion value ≈ 6 %.
 245

Second Case: Inside Track (2.38 inches) and 4 kHz

Using the same equations as before with value changes 𝑎 = 0.00009 𝑎𝑛𝑑 𝜆 =
0.00498.

For 3.75 mil, 𝐾𝑟 = 4.73 𝑎𝑛𝑑 𝐾𝑎 = 0.11 which gives from figure 22-2 a total
distortion value ≈ 20 %.

For 0.7 mil, 𝐾𝑟 = 0.88 𝑎𝑛𝑑 𝐾𝑎 = 0.11 which gives from figure 22-2 a total distortion
value ≈ 5 %.

Other calculations can be performed on the outside track and for other stylus values,
however, conclusions can be drawn for the best stylus to transcribe Edison Diamond Disc
Records. The use of a current DJ style 0.7 mil spherical shape will perform better than the
original Diamond Disc stylus and should be used. The 3.5 mil Stylus used by Edison was
needed to support the heavy weight in order to produce a reasonable acoustic output using
strictly mechanical means. The fact that a smaller radius for the stylus would provide a better
reproduction of the sound was understood by Edison and a patent from 1900 numbered
652,457 describes the elliptical style of stylus that is used today. My understanding of the use
of the larger 3.5 mil spherical stylus by Edison was to support the heavy weight needed for
practical use of a mechanical sound system.
Because the stylus rides on the bottom of the groove for vertical recording, a smooth
and polished tip (as shown in the Edison Drawings) is required. The DJ style seems to have
a smooth tip even though this stylus is designed for the LP records where the stylus does not
ride on the bottom. Figure 22-8 was provided by the Ortofon Company for their Reloop OM
Black DJ Style of stylus. The stylus is a spherical shape and will ride with full contact into
the cut groove of a diamond disc since the groove shape is also a spherical shape.
 246

Figure 22-8 Ortofon Stylus Information

18 micrometers correspond to 0.7 mil and the groove shape in the drawing is for lateral
modulation (4 micrometers is the shape of the groove bottom) application. Ortofon customer
service has stated that the bottom and sides of the stylus are polished.
Another very important benefit of using this 0.7 mil tip is that the shape of the noise
that remains contains high frequency energy which in turn provides the DCart noise removal
algorithms information to remove noise events. When similar records using the larger 3.75
mil stylus are processed with the same noise filters settings, more noise events remain versus
using the 0.7 mil stylus.

22.5 Stylus Selection for 78 RPM Records

22.5.1 Introduction

The original stylus used for playback of 78 RPM records was a soft steel shaped point
that was designed to wear into the record groove within a couple of record revolutions. Steel
 247

particles and years of dirt will often be found in the 78 RPM record grooves even after a good
cleaning. Unlike the Edison Diamond Disc records, the stylus must ride above the bottom of
the groove. If the stylus is in contact with the bottom of the groove, you will hear a large
increase in the noise.
After many record playbacks using the older reproducers, a significant amount of wear
in the groove walls will exist which adds to the background noise. The stylus selection for 78
RPM records requires a shape that does not ride on the bottom and a width that is in contact
with the smoothest part of the record groove. The best approach is to have a selection of stylus
sizes to experimentally find the best transfer since each record has a unique groove regarding
the location of wear.

22.5.2 Worst Case Calculations for Two Stylus

Experimental results found that a 3.5 mil spherical shape stylus used with the Nagaoka
MP-110 cartridge provided a good transfer for very worn 78 RPM records where the original
designed stylus used a radius of about 2.5 mil.
Let’s compare the 3.5 mil stylus and the 2.5 mil version for worst case conditions at
the inner radius using 3 kHz and 5 kHz music waves. The equations come from a similar
analysis based on 22.3.1.

First Case: Inside Track (2.0 inches) and 5 kHz sound wave. Using the graph from
figure 22-3 and equation 22-9, a 2.0 mil radius gives 20 % total harmonic distortion and a 2.5
2
mil stylus is 2.5 ⁄ 2 = 1.56, 1.56 × 20% 𝑜𝑟 ≈ 31 % . Using similar calculations, a 3.5
2.0
2
mil stylus gives total harmonic distortion of 3.5 ⁄ 2 = 3.06, 3.06 × 20% 𝑜𝑟 ≈ 61 %.
2.0

Second Case: Inside Track (2.0 inches) and 3 kHz sound wave. Using graph from figure
22-3, a 2.0 mil radius gives 8 % total harmonic distortion and a 2.5 mil stylus ≈ 12 %. For a
3.5 mil stylus and again using figure 22-3 the total harmonic distortion is ≈ 24 %.

When comparing these numbers with those of the Diamond Disc recall that a larger
amplitude of the sine wave has a squared effect on the lateral distortion and that the Diamond
Discs used a very low amplitude versus the lateral records. The result from using the graph
shows that the best stylus to minimize the distortion from the recording-playback process is
to have a small radius that does not ride in the bottom of the groove. With the worn condition
 248

of many 78 RPM records, a larger radius helps to reduce the noise from the surface with the
tradeoff of more harmonic distortion on the high frequencies.

22.6 Conclusion

The mathematical analysis presented in the book “Sound Recording” by Frayne and
Wolfe which built on the original work from Pierce and Hunt provided a useful mathematical
relationship between the stylus radius and distortion of the recorded sound groove for both
Lateral and Vertical recording. Because information is available from the Edison Music
Company regarding actual diamond disc recording values, calculations were shown that a
current, DJ style stylus of 0.7 mil radius will provide a low value of distortion for the Diamond
Disc Records. This small radius can be used because the stylus rides on the bottom of the
groove where the sound is located. This is in sharp contrast to the 78 RPM records that use
lateral motion from the groove walls.
The 78 RPM records cannot use a stylus that rides on the bottom of the groove with
gives a minimum radius size. Because the record groove had to support and move a heavy
reproducer and shape a soft steel needle, the condition of many of the groove walls have worn
sound shapes. Thus, a stylus that rides on a clean part of the groove wall may provide the best
result even though high frequency distortion will result.
The original phonograph patent that Thomas Edison wrote claimed vertical and lateral
forms of groove motion to record the sound. As time continued from the initial invention, two
different music industries developed, one by Edison with the vertical motion and another
started by Emile Berliner using the lateral motion. The lateral motion on a flat surface
provided several early advantages for Berliner by providing a groove that could propel the
tonearm by itself and provide a method to reproduce copies by making a metal stamping
mold. Edison continued to develop the vertical recording method and needed to add
mechanical features to move the tonearm without energy from the sound grooves.
The fact that the sound is placed into the groove using a chisel shape for both vertical
and horizontal groove motions and a finite radius to the stylus is used to follow or trace the
groove motion means that there will always be some distortion to the signal. The lateral
motion with sound on the walls requires a minimum radius to ride higher than the bottom in
contrast to the vertical motion that can have a small radius ride on the bottom of the groove.
Since low mass tonearms and electric amplification is available, a much smaller stylus
radius can be used for the Edison Diamond Disc records than was originally needed (also the
mechanical driven lead screw is not required since the groove can move the low mass
 249

tonearm). The result is that for the Edison Diamond Disc Records a lower distortion to the
music is possible versus the 78 RPM records. The 78 RPM records will always require a
larger stylus radius than the Diamond Disc Records to ride above the groove bottom. The
resulting distortion to the music has been shown to be significant between these two methods.

Edison Believed that His Recording Method was Better than the Competitions for
Recording sound. Today this belief can be realized, and we can use this to improve the
Sound that was originally heard on the Edison Phonographs.
 250

23.0 Audio Standards

This chapter describes standards that have been developed to measure signal levels.
These standard measurement methods provide interoperability between various audio devices
when they are connected to form the needed audio system. This connecting of devices to each
other requires that the electrical levels are compatible and predictable so that when connected
they will achieve the desired result. For example, when a CD player is connected to a power
audio amplifier and then a set of speakers you expect that all the parts will work together. The
volume control on the power amplifier should give a smooth increase from low to high and
not start off at a very loud level and increase to a high distortion. This level of expectation
requires that the inputs and outputs of the audio equipment use some sort of standard values
for audio levels so that different devices work well together.

23.1 Logarithms, a needed Concept

The way that the human ear perceives changes in pitch and amplitude is not linear but
is logarithmic. The uses of logarithms are useful to describe hearing with the ear.

The logarithm of the ratio between two powers is expressed in bels:

𝑃1
𝑏𝑒𝑙𝑠 = 𝑙𝑜𝑔10 𝑒𝑞(23 − 1)
𝑃2

Where the bel is a ratio and by itself has no absolute value without a reference for the
0.0 value. 𝑃1 and 𝑃2 are signal powers.
The bel is too large a value for most audio work so a new term, the decibel (dB), is
used which is a tenth of a bel. Therefore:

𝑃1
𝑑𝐵 = 10 ∙ 𝑙𝑜𝑔10 𝑒𝑞(23 − 2)
𝑃2

When the ratio uses voltage or some other value instead of power another change to
the formula is needed. For voltage a squared term is used since power equals voltage squared
divided by the real part of the impedance. Therefore, for voltages the equation becomes (using
log properties):
 251

𝑉2
𝑑𝐵 = 20 ∙ 𝑙𝑜𝑔10 𝑒𝑞(23 − 3)
𝑉1

Where 𝑉1 and 𝑉2 are Volts and 𝑉1 is the reference.

Some examples of the use of dBs in sound (for most people):

1. The minimum change in sound level that can be heard is 1 dB.

2. The threshold of hearing is 0 dB.
3. For the intensity of sound to double the change needed is 3 dB (independent of the
specific level).

Example three is very interesting. Let’s say that the amplifier is putting out 100 watts
to a loudspeaker and you want to double the intensity of the music. The needed power to
double (3dB) the intensity will be 200 watts. Next, you want to again double the intensity.
Now you will need 400 watts. This relationship is not a linear one rather it is logarithmic.
While logs were originally invented to help with multiplying numbers, they turned are useful
because of their relationship to the way that the human ear (and vision) responses to changes.

23.2 Early Sound Intensity Measurements, the VU Meter

Sound intensity measurements have been needed since the study of acoustics began.
An early method to measure sound was a Rayleigh Disk (circa 1882) that would rotate in
accordance with the impinging sound wave. The amount of rotation of the disk indicated the
strength or intensity of the sound. As audio technology improved (1920’s to 30’s), the various
companies developed standards for the volume level of sound to allow the audio components
to work correctly within a complete system. Initially the work was not unified, and each
company and industry used their own methods so that telephone, radio, movies, and record
companies all used unique methods for measurement and standard levels.
Late in the 1930’s an effort was made by three engineers named H.A. Chinn, D.K.
Gannett and R.M. Morris to develop an industry standard Volume Indicator and Reference
Level device for the audio industry. The well-known VU meter and the associated circuitry
was the result of their work. This VU meter and the resultant 0 reading served as a reference
value and became widely used within the audio industry. This meter is still made today as
either an actual physical meter or as a digital device with a software display. The dynamic
 252

ballistics of the meters’ mechanical movement was required to meet a specific set of values
so that the indicated value would be useful for the complex audio wave shapes to be measured.
To understand the reason for the meters’ mechanical specification, start with a sine
wave at an audio rate of 1 kHz. Refer to the figure 23-1:

Figure 23-1 1 kHz Sine Wave

For this sine wave, the P-P (peak-to-peak) value and the RMS (Root Mean Square)
value can be related to each other with a simple formula. The P-P value is the amplitude from
the lowest to highest value and the RMS value is the P-P value divided by 2√2. The power or
intensity of this audio wave form is then the RMS value squared divided by the Real part of
the Impedance of the load that this wave is applied to.
For music and voice pure sine waves do not occur. Let’s look at an actual wave form
from a song that is shown in figure 23-2:

Figure 23-2 Song Wave

To calculate the P-P and RMS values of this song you will need to specify where you
perform the measurement. In addition, you will need to perform actual integration of the
waveform (area under the curve) to measure the RMS value. The term used for sine waves of
√2 is not valid for any shape, just for a sine wave. Since the P-P value is needed to avoid
overdrive or saturate of electrical devices and the RMS value is needed to measure the
 253

effective power or intensity of the sound wave some type of device is needed that is useful
for these complex audio waveforms. This need for measurement of audio complex waveforms
was the reason that the VU meter was designed and had specific mechanical ballistic
properties. The response of the meter’s pointer to a complex signal (mechanical ballistic
properties) was designed to find a value that was representative of the volume or RMS value
of the signal as best that could be for a wide range of different audio waveforms. This meter
could then be useful for all types of songs with instruments and vocals. The meter had a
calibration point where 0 on the scale corresponded to one milliwatt of power with a load of
600 ohms (resistive). The other values on the scale were in dB units above and below this
reference point. The value of 1 milliwatt into a 600-ohm (resistive) load as a dB reference is
known as 0 dBm. This reference value corresponds to 0.7746 V RMS for a sine wave.
The VU meter was used within the audio studio to allow the music producers to set the
levels for the different musician’s microphones so that the desired balance between the
musicians could be made by watching the meter while setting the gains of the audio amplifiers
rather than setting a balance by just listening. The VU meter had many other uses in radio
broadcasting to provide a useful measurement of the audio sent to the transmitting circuits.
Along with the VU meter additional standards were developed for the new audio
equipment available. Note: The DCart10 software provides an audio level meter that is useful
but does not simulate an original VU meter in terms of the meters mechanical movement
(ballistics) or 0 dB calibration values.

23.3 Standards Important for Audio Restoration Work

Currently audio equipment uses a term for signal levels called line level for both inputs
and outputs. In general, the line level inputs for converters have a nominal value of -10 dBv
for consumer audio. The professional audio line level nominal value is +4 dBu.
All dB numbers need a reference since the calculation of dB must have some definition
for 0.0 dB. In the case of the dBv value the 0.0 dB reference value is a signal of 1.0 V rms.
Therefore, the value of -10 dBv is 0.316 V rms or 0.894 V P-P for a sine wave. When power
is converted to voltage with a 600-ohm load, 0 dBm = 0 dBu = 0.7746 V rms. Therefore,
+4dBu = 1.228 V rms or 3.473 V P-P.
The term nominal can be confusing since the difference between nominal and max
(headroom or when saturation occurs) is not often stated by the converter manufacturer.
Another term that represents the maximum signal that the converter can accept (without
distortion) is known as 0.0 dBFS since the digital output will be all 1’s at this point. For
 254

example, if the specification said that the peak level: 0 dBFS @ +6dBv that means the
maximum output for the A to D converter will occur for an input signal with an RMS value
of 2.0.

23.3.1 Head Room and Alignment Levels

As we saw in figure 23-2 audio waveforms have complex shapes and have a large range
from loud to soft passages. The signals complex nature caused the development of VU meters
in the past to find a typical or nominal level that represented the most likely power value.
Today with digital recording being used, the audio industry uses a phrase called alignment
level to represent the 0 value on the VU meter scale or the consumer phrase nominal level.
The concept is that the alignment level is set to a value that allows adequate headroom or
space from the typical level of the music to the maximum 0.0 dBFS value; in this manner the
music will not overdrive or exceed the maximum digital value for the digital word length
used. The range for alignment level to maximum is around 20 dB for most audio commercial
systems.
For the A to D and D to A converters that are available; most of them just state that
they have a line level input and do not tell you what the 0.0 dBFS value is. Therefore, the
headroom is not specified. Some of the converters used for professional audio use do provide
the needed information.
Since the DCart10 software provides meters, we have a way to observe the signals
amplitude and see if it is over-driving or exceeding the 0.0 dBFS value by using the vertical
amplitude scale or the DCart10 version of a VU meter.

23.4 Diamond Cut Software Level Indicators, VU and Vertical Axis

For the DCart10 software the algorithm receives the output from the converter without
any direct information regarding what a maximum (all 1’s) means. When the converter sends
a maximum digital value the actual signal level will depend on its design. The DCart10
software assigns a value of 0.0 dB to represent the maximum value from the analog to digital
converter without a direct linkage to 0.0 dBv or 0.0 dBu. The DCart10 software does not
know what has been defined for a maximum absolute value. All that the software knows is
that the maximum value will be all 1’s and will define 0.0 dB as that value.
For the VU scale that is displayed on the right side of the window the maximum
amplitude has a scale value of 0.0 dB at the top of the software meter. The green line is the
 255

highest value that has occurred in the song and the white line is a filtered or averaged
response. The software meters in the DCart10 software do not correspond to the previously
defined ballistic mechanical properties of the original VU meter although there is an
adjustable marker reaction time (set in the preferences section). The general term VU is
displayed by the software but should not be taken to mean that the meter represents the
properties from the late 1930’s VU meter definitions. The scale can be changed from a display
with two scales called linear or log. Note that in both cases the logarithm of the amplitude is
shown (0.0 dB is full scale). The terms linear and log refer to the spacing between the values
not the method of calculating the values. In all cases the dB of the amplitude with respect to
a full-scale value of 0.0 dB is shown.
The display of the waveform versus time has an adjustable vertical scale (set in
preferences) that can show the output from the converter in digital values or a conversion to
dB. For a linear output scale use the counts option.
Figure 23-3 shows a picture while a song is being played with a linear scale using the
digital value from the converter. Figure 23-4 shows the same picture with a log scale.
The best method to locate and remove noise from the music (figure 23-3) uses the
amplitude scale displaying counts.

Figure 23-3 Amplitude Scale for Waveform

 256

Figure 23-4 contains the same waveform as in Figure 23-3 with the vertical scale
changed to dB.

Figure 23-4 dB scale for Waveform

 257

24.0 Correct Recording Level & Number of Bits to Use

This chapter will examine the optimum level and number of bits needed to record the
music as digital values in your recording studio. Key concepts will be shown that relate the
effective number of bits in the A to D conversion to the selected number of bits.

24.1 Details of The Analog to Digital Conversion

The signal from the preamplifier, being an analog signal, is continuous with no specific
discrete values. When this analog signal is converted to discrete digital values that conversion
operation (A to D) provides a digital value that is close to but not an exact match to the actual
analog value. How close the digital value is to the analog value depends on how many bits
are available to represent the input and how much noise or uncertainty is present in the
conversion. Since the digital value is limited to a finite number of bits the best that can be
done in the conversion is to have the conversion accuracy within ± one LSB or least
significant bit. For example, if the input to the A to D ranges between ± 0.5 Volt or 1.0 V full
range and there are 8 bits in the word length, then the LSB is:

𝑓𝑢𝑙𝑙 𝑣𝑜𝑙𝑡𝑎𝑔𝑒 𝑟𝑎𝑛𝑔𝑒 1.0

𝑙𝑠𝑏 = = 8 = 3.91 𝑚𝑖𝑙𝑙𝑖𝑣𝑜𝑙𝑡𝑠 𝑒𝑞(24 − 1)
2𝑤𝑜𝑟𝑑 𝑙𝑒𝑛𝑔𝑡ℎ 2

This equation shows that the conversion from analog to digital has a finite ability to
get an exact value; the conversion will have a potential error of up to ± 3.91 millivolt. While
that value is small, it is not zero. As the number of bits is increased the conversion becomes
closer and closer to the actual value.
When digital values are used to represent an analog value, a method is needed to
indicate that the voltage had a positive or negative value since the analog signals are AC
(alternating voltage) and their value swings positive and negative. A common method used
to indicate polarity is called two’s complement format which uses the MSB (most significant
bit) to indicate polarity. Using the two’s complement format, the largest positive signal
is(2𝑏−1 − 1) and the largest negative signal as (−2𝑏−1 ) where 𝑏 is the number of bits in the
digital word. The complement part of the format refers to additional bit changes to the
negative word to make future math operations less complicated in hardware. There are other
methods used to indicate the polarity of the digital word, but they also use a part of the digital
 258

word for this purpose and therefore the number of bits that represent the analog signal is
slightly less than the total number you start with.
At this point you may believe that the best approach for transcribing music would be
to use the maximum possible number of bits available for the analog to digital conversion
process. A practical limit is defined as the effective number of bits (ENOB) in the conversion
which occurs with the presence of electrical noise or analog signals that are not related to the
music. Adding bits greater than the ENOB is a waste of system resources and adds no value.
A useful concept in Audio Restoration to understand this limitation will be the term Signal to
Noise Ratio. This term is used in many other fields of engineering.
As the equations for the signal to noise ratio are developed, they will show how the
noise and the effective number of bits interact with each other.

24.2 Signal to Noise Ratio (SNR) for Audio

The Signal is the desired value we want, and Noise is anything extra that is present
when the signal is applied to a device.
The Signal Power to Noise Power, as measured in Decibels, is defined as:

𝑃𝑜𝑤𝑒𝑟 𝑆𝑖𝑔𝑛𝑎𝑙
𝑆𝑁𝑅𝑑𝐵 = 10 ∙ 𝐿𝑂𝐺10 ( ) 𝑒𝑞(24 − 2)
𝑃𝑜𝑤𝑒𝑟 𝑁𝑜𝑖𝑠𝑒

Using the Statistical values of the signal, the relationship can also be written as:

𝜎 2 𝑠𝑖𝑔𝑛𝑎𝑙
𝑆𝑁𝑅𝑑𝐵 = 10 ∙ 𝐿𝑂𝐺10 ( 2 ) 𝑒𝑞(24 − 3)
𝜎 𝑛𝑜𝑖𝑠𝑒

Where 𝜎 2 is defined as the Variance of the signal and 𝜎 is the RMS or one sigma of
the signal. This equation can be further modified using logarithmic relationships and with an
assumed load of 1.0 ohm:

𝑉1
𝑆𝑁𝑅𝑑𝐵 = 20 ∙ 𝐿𝑂𝐺10 ( ) 𝑒𝑞(24 − 4)
𝑉2

Where V1 is the voltage of the Signal and V2 is the voltage of the noise.
 259

24.3 A to D Noise

There is a source of noise within the A to D conversion that occurs from the uncertainty
in the LSB or quantization error. This is not the same as external noise within the system;
rather this is noise due solely to the analog data being converted to distinct digital data values.
Another source of noise within the A to D conversion involves the relationship between
the input signals shape and the range of the conversion process. Here the shape refers to how
much of the range of the A to D conversion is used for the bulk or majority of the signals
values as a comparison to the potential or full range of the converter. For example, if the range
of the A to D converter is 2 volts and the signal is only 25 millivolts P-P in value, then much
of the converter’s range is not used. The opposite of this is the case where the signal uses the
total converter’s range with all the possible digital values produced. An electrical term to
define and measure this is called LF (Load Factor) and will be defined as:

𝑟𝑚𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑝𝑢𝑡 𝑠𝑖𝑔𝑛𝑎𝑙 𝜎 𝑠𝑖𝑔𝑛𝑎𝑙

𝐿𝐹 = = 𝑒𝑞(24 − 5)
𝑉𝑝 𝑉𝑝

Where Vp is the converter’s peak voltage and 𝜎 𝑠𝑖𝑔𝑛𝑎𝑙 is the one sigma-value of the
signal.
After some mathematical work and including the Signal Power to Noise Power
(decibels), the relationship showing how quantization (converting an analog signal to a digital
value) and signal shape works to define the digital word bits is:

𝑆𝑁𝑅 = 6.02 ∙ 𝑏 + 4.77 + 20 ∙ 𝑙𝑜𝑔10 (𝐿𝐹) 𝑒𝑞(24 − 6)

Where b is defined as the number of bits in the converters digital word.

This equation is often seen in a form with the assumption that a sinewave is the sole
input to the A to D converter and that the peak of this signal is the same value as the peak
converters maximum value. In this case, the equation reduces to:

𝑆𝑁𝑅 = 6.02 ∙ 𝑏 + 1.76𝑑𝐵 𝑒𝑞(24 − 7)

At this point we have an accurate relationship (equation 24-6) for the analog to digital
converters performance related to a signal to noise ratio. The equation gives important
information relating the Signal to Noise to the number of bits used to digitize the analog
 260

signal. We would want the SNR value from the converter to be much larger than the audio
signal SNR value so that we do not add extra noise to the music during the recording process.
Some examples of the SNR values using the simplified equation (24-7) are:

1. 8 bits yields 49.92 dB

2. 12 bits yields 74 dB
3. 16 bits yields 98.08 dB
4. 24 bits yields 146.24 dB

These values, while a starting point, are not accurate in a real audio system. In the next
section a relationship will be developed that shows how the noise in the system further
degrades the actual or effective number of bits for the analog to digital process.

24.4 Measuring Signal to Noise Ratio and Effective Number of Bits

When any noise is present in the analog to digital system the ability to use all the digital
bits is affected. For example, if the electrical noise is greater than the input analog signal, then
the conversion to a unique digital value will not happen as the noise signal will add to or
subtract from the desired value. Noise that occurs in the recording system comes from many
different sources. These sources are often not related to each other or uncorrelated with each
other. The typical audio analog signal does not have noise present that will be occurring in
exact timing with other noise in the preamplifier or the converter. Before an equation can be
developed a modification to the SNR term is needed to reflect how the noise powers and
signal powers interact. The concept is to have a term that can be measured in a real system
that has a signal present along with noise and distortion. A new term labeled SINAD will be
used with stands for signal to noise and distortion ratio.

The equation for this term is:

(𝑃𝑠𝑖𝑔𝑛𝑎𝑙 + 𝑃𝑛𝑜𝑖𝑠𝑒 + 𝑃𝑑𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛 )

𝑆𝐼𝑁𝐴𝐷 = ( ) 𝑒𝑞(24 − 8)
(𝑃𝑛𝑜𝑖𝑠𝑒 + 𝑃𝑑𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛 )

Where 𝑃 is defined as the power of a variable.

 261

Another term called the ENOB, or effective number of bits is used to define the effect
of noise on A to D performance. The ENOB value tells you what the real or effective number
of bits is, verses what the converter is designed for. ENOB is the result when the system is
used with noise present. The ENOB is always less than the original number of bits in the A
to D converter and at times can be very much less.

A common equation for the ENOB is:

𝑆𝐼𝑁𝐴𝐷 − 1.76
𝑏𝑒𝑓𝑓 = 𝑒𝑞 (24 − 9)
6.02

Where 𝑏𝑒𝑓𝑓 is defined as the effective number of digital bits.

This equation is the same as equation (24-7) when solved for digital bits and SINAD
substituted for SNR. This equation shows that the effective number of bits (beff) is equal to a
number that is dependent on the SINAD or how much noise and distortion is present and not
the actual number of bits in the A to D converter. The equation is not a firm mathematical
relationship but rather a general expression that demonstrates that the presence of noise in a
system will have a significant effect on the actual conversion to a digital value.
The audio recording system that is used to transcribe the music uses an A to D converter
that will offer a variable number of bits in the digital word to represent the analog value. This
equation can determine the actual number of bits that are used to convert music on a record
to a digital word for an actual audio system.

24.5 System Example of SINAD

The DCart10 software, along with an Excel Spreadsheet program, provides a method
to measure the Effective Number of Bits on a recording system. The audio system described
in Chapter Four will provide actual values.
The first step is to define where the Signal to Noise measurement will be performed,
as there are three main locations for the noise measurement.

1. The Total recording system from A to D up to and including the record cartridge.
2. The equipment from step 1 and including the noise from the record groove wall (surface
noise).
 262

3. The result from 2 and additional noise from the recording music session.

The first step can be measured and is a constant for all the music recordings that you
perform as this noise is always present. In addition, if we perform some measurements on
value one; we can then get an idea on how many bits we should use for our recording. An
example is shown in figure 24-1. This was previously used in the transcribing chapter as a
raw recording.

Figure 24-1 Quiet before Music

Prior to the start of the music there is a time where the DCart10 software has started
recording and the cartridge stylus has not yet touched the record and started playing music.
This location has a marker labeled quiet in figure 24-1. Later the music has started, and that
area is shown with the marker labeled music. The markers are in the general area where the
music and quiet section are located; they do not show the actual boundaries of these events.
The beginning section, while quiet, does have some noise present. To measure this
noise (which is the system noise with no signal present) the section is highlighted and copied
to a new file with the DCart10 software. The result of this copy is displayed in figure 24-2:
 263

Figure 24-2 Quiet Section (System Noise)

While the values in this section appear to have no value (straight line) they do have a
small number of counts coming from the converter. They cannot be seen in this figure due to
the vertical scale that is used for display. Notice that the largest and smallest value shown is
± 32 K. The total range of 64 K represents the 16 bits used to record this music.
Using the DCart10 software this section can be saved as a csv file and then imported
into a spreadsheet for statistical calculations. In a similar manner the music section that
contains the recording and noise can be saved as a csv file and imported into a spreadsheet
for calculations. These files contain many data points due to the sampling rate (96 K per
second). Some spreadsheets impose limits on the number of data rows and columns that can
be entered and therefore a sample of the quiet and music time section may have to be used
instead of the whole section. Using the math tools in the Excel Spreadsheet program, the
results are:
1. The Variance of the Signal section was 0.0010822193 counts.
2. The Variance of the Quiet section was 0.0000000387 counts.

Using equation (24-8) with the correct terms:

𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑠𝑖𝑔𝑛𝑎𝑙 + 𝑛𝑜𝑖𝑠𝑒 0.0010822193

𝑆𝐼𝑁𝐴𝐷 = 10 ∙ 𝐿𝑂𝐺10 ( ) = 10 ∙ 𝐿𝑂𝐺10 ( )
𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑛𝑜𝑖𝑠𝑒 0.0000000387
= 44.46𝑑𝐵

Now, using equation (23-9)

(44.46 − 1.76)
𝑏𝑒𝑓𝑓 = ≈7 𝑒𝑞(24 − 10)
6.02
 264

Even though we started with 16 bits in the converter we really have about 7 bits of real
digital information representing the music with the lower 9 bits just representing noise. The
7-bit value is an approximation but if we started with 16 bits for the converter, we would have
more than an adequate range for this music. It’s important to remember that the calculation
for SINAD uses the raw recorded music before noise removal and normalization by the
DCart10 software. The final SINAD result could be better with some noise removed.
Using another song that had much of the noise removed and was a very good recording
yielded a variance value for the Signal of 0.0048001210 which after using the same equations
produced a value of 𝑏𝑒𝑓𝑓 ≈ 8 . The use of SINAD in this manner, while not perfect, gives us
insight into how many bits should be used for the transcribing of music on our home system.

24.6 Conclusion

The calculation of SINAD is a useful concept that should be used with care. If the
literature for audio recordings is searched for signal to noise of records the reported
performance can range from as low as 30 dB for early and worn 78 RPM recordings up to as
high as 60 dB for special vinyl records used in transmitting music in radio stations. The value
of 44 dB that was measured in the experiment is within the general range for audio recordings.
Therefore, the noise from the record itself has a large influence on the specific SINAD
calculation.
The general subject of audio noise and how to measure it is more complex than the
examples shown here. There are many additional measurement methods and techniques that
have been developed in this audio field that allows a precise measurement of SINAD and the
resulting ENOB. For example, the Load Factor (LF) could be measured rather than assumed
to be the best that could be used since the simplified equation (24-7) was the basis of the
ENOB and an actual measured value for LF would decrease the ENOB value.
However, for our needs in audio restoration the answer for the number of bits to use is
direct and relatively simple; namely the use of 16 bits for the word length of the digital
conversion will provide a potential SNR of 94.56 dB (equation 24-7) and since the typical
recording system has noise from the record as the significant term the SINAD is much less
than the potential value and the resulting ENOB value of 7 much less than the starting 16
value.
The actual gain value to use in the transcription uses equations (24-5) and (24-6) to
demonstrate that the Load Factor should be as large as possible which will reduce the SNR
 265

number of the A to D process to the minimum possible amount. The Peak Voltage or V p that
the converter uses is a fixed value while the one sigma or RMS value of the music can be
varied by using the preamplifier gain control. You can set the gain so that a small amount of
head room of about 3.0 dB remains which lets the range of the song extend over a wide range
of the A to D process. Thus, the loudest parts of the music should be around 3.0 dB below the
maximum value and the average value should indicate around 20-10 dB down. The average
value will vary quite a bit, but the idea is to have the RMS of the music (one sigma) as large
as possible. The noise spikes in the music from defects and wear will often range high enough
to show a red meter reading indicating that the signal has exceeded the maximum range. The
fact that the noise spikes saturate the converter is in general O.K. The DCart10 software can
remove them whether they saturate or not the A to D converter.
 266

25.0 Providing the Needed Preamplifier Gain for Transcription

25.1 Introduction

The analog signal produced by the phonograph cartridge is very low in amplitude while
the music is being played. This signal needs to be increased as it’s fed into the line-input on
the converter. The question for restoration work is how much the raw cartridge signal should
be increased as it passes to the A to D converter.
The needed amplitude increase of the signal from cartridge to A to D Converter can be
calculated in the following manner. First the desired result of the creation of the digital music
file is to have the best or largest Signal to Noise Ratio. The previous chapter covered this in
detail. This chapter will determine the amount of overall gain that is needed for a music signal
to produce more than the maximum A to D output.
The example will use the audio system in chapter four and can be adapted for your own
audio system. For the amplitude of the signal or S value, the needed gain is to have the ability
to always exceed the maximum input of the converter. The 0.0 dB level (maximum) for a
typical converter is about 1.0 Volt P-P (Peak to Peak) and would correspond to the maximum
digital value. Therefore, the preamplifier’s gain should be able to increase the output from the
cartridge, for all records, to a value exceeding 0.0 dB as shown on the software VU meter.
The best practice is to have additional gain control past the 0.0 dB value so that a wide range
of record levels can be played.
For a moving magnet Stanton 500.V3 cartridge (a common type), the stated output is
4.6 mV (P-P) into a 47 K ohm load while playing back a 1 kHz tone. Much of this
specification depends on the specific type of recording; however, the 4.6 mV value provides
a useful starting point.
To increase the 4.6 mV output to 1.0 V, a gain of 217.39 is needed. This gain value can
be also be represented as ≈ 47 dB. The actual gain will not be exactly 47 dB for all conditions
since the recording and transfer from the record to a digital file will involve a compromise to
keep the signal to noise ratio at the optimum value.

25.2 Preamplifier Gain Values

Most of the commercial preamplifiers do not state how much gain the device provides.
In addition, most preamplifiers do not offer a variable gain control rather a fixed gain in the
 267

internal circuits. The expected operation by the user is that the gain can be varied by using
the software controls on the converter or a 24-bit recording converter range is used which is
large enough to cover all possible values. This large range allows the signal from any record
to be amplified, in a digital conversion, to restore the music. While this concept may work, it
is not the best. When a signal is low, and it is amplified any noise that is present is also
amplified by the same amount. The Load Factor of this conversion process is low and as we
saw this will increases the system noise. The best method is to have a large signal swing
during the initial transfer, and this will in turn provide the best signal to noise ratio for the
restoration work.
Two different methods are used in my Music Studio with excellent results. Both
solutions use a fixed gain flat preamplifier (DCP-47K-F from Diamond Cut Productions).
The first solution was to send the preamp output to a Behringer analog mixer (Xenyx 502)
with variable gain on both each channel and an overall value. The mixer output then fed into
the line-in port on the Realtek Chip Set A to D converter. The flat preamplifier from Diamond
Cut Productions had been modified to provide a fixed gain of 26.52 dB. This value was chosen
so that the output of the flat fixed gain preamplifier would match the expected input level of
the Behringer mixer device thus providing a large range of gain control.
A second solution was to use a variable gain A to D converter from the Focusrite
company (Scarlett 2i2 Gen3). The same fixed gain flat preamplifier provided the input to the
Focusrite Converter. The Focusrite Converter features adjustable gain on each channel. This
variable gain A to D converter, when used with the flat preamplifier, provided an alternative
to the Behringer mixer and the Computer Realtek Chip Set.
Both methods provide the needed extra gain (more than 20 dB) for the music
transcription. Using the Behringer mixer allows the use of other analog signals besides the
turntable, to be converted to digital values.
 268

26.0 Sample Rate Needed for Restoration

This section will cover the subject of what the sampling rate should be for audio
restoration for both the accurate reconstruction of the signal and needed DSP (digital signal
processing) in the software.

26.1 History of Sampling Analog Data

When the subject is: “What is the correct sample rate to accurately convert an analog
signal”? The first thought is that the rate should be at least twice the highest frequency. While
this comment is true it needs some additional understanding to make sense when restoring
music. Some early history into sampling theory will help provide answers.
An early pioneer of Radio was Reginald Fessenden. Fessenden developed several
major principals in Radio during the beginning of the 20 th Century. He developed the
heterodyne principal and has been credited with making the first broadcast of the human voice
over radio waves. This first broadcast for him was in the fall of 1906 from a station in
Massachusetts. At this time, the phrase Wireless Telephony was used to describe what we
today call AM (Amplitude Modulation) Radio. His work was performed without using any
vacuum tubes or solid-state devices. To generate the high frequency RF energy for
transmitting over the air, Fessenden used a rotating spark gap generator to produce a series of
impulses to tuned circuits. The RF energy was modulated to produce AM threw the use of a
water-cooled carbon microphone while he spoke and played music. He wanted to understand
the relationship between the speeds of the interruptions by the rotating spark gap device to
the audio frequency that was being transmitted. At the time of his experiments, he determined
that for voice frequencies a minimum frequency of interruption of 10,000 per second was
needed for good fidelity, however, because the same interruptions were used to produce the
RF energy, he used a higher rate up to 25,000 per second to make an AM RF signal. He stated
in an early report that what was really needed was a RF generator that produced a continuous
waveform that would be in turn modulated by the audio frequencies. The continuous
waveforms were later developed by using rotating high-speed alternators (over 100,000 Hz),
arc spark gaps and then the vacuum tube. What is interesting is that this early work with the
rotating spark gap was in effect an early sampling device and while voice frequencies are at
most 3000 Hz the value, the value found of 10,000 Hz is not too far away from the common
8000 Hz number used currently in telephone circuits. Fessenden presented a paper on the
 269

history of Wireless Telephony to the Smithsonian Institution in 1908 that provided this
information, and that paper is a wonderful source of Radio information.
In 1922, another Radio pioneer name Edwin Armstrong invented a modification of his
regenerative patent that improved the performance of this device by interrupting the
oscillation in the detector stage at a high audio rate around 20,000 Hz. He called this new idea
Super Regeneration. Armstrong had previously invented Super Heterodyne for radios and
would go on to invent FM (Frequency Modulation) Radio.
By interrupting the regenerative feedback, the circuit was able to increase the gain of a
single tube to over one million. Today this same circuit is used in ultra-high frequency circuits
to provide reception with high gain and low cost. At the time that Armstrong invented the
circuit he specified that the interrupting frequency had to be about two times the highest audio
signal to work. It isn’t clear from reading his patent for super-regeneration if the mathematical
principals for sampling theory were understood by him at the time, but his factor of 2 was
correct.
In 1928, Harry Nyquist published a technical paper, “Certain Topics in Telegraph
Theory” that was presented at the Winter Convention of the A.I.E.E. The paper listed several
topics centered on the technical aspects of transmitting code signals over wires for the
important telegraph industry. This paper describes many principals of digital communication
that have important uses today with wireless transmissions. These important principals would
be built upon by many more scientists to provide the needed technology that we use for our
wireless communications. One of the areas that Nyquist discussed in this paper was the
relationship between the bandwidth of the electrical circuits that transmit the signal and the
frequency content of the pulse that is to be sent. A careful reading of his paper does not reveal
all the needed description of a complete sampling system that converts an analog signal to
digital and then back again to an exact analog signal. Rather, the work by Nyquist was built
upon by Claude Shannon and others to create the Sampling Theorem that we use today for
analog to digital and digital to analog conversions. However, the phrase “Nyquist Sampling
Rate” is often used to encompass the work started by Nyquist and later developed by others.

26.2 Sampling Theory

Sampling Theorem details are available in many communication references today for
the reader who wants to dig deeper into the theory. The core of the idea is that if you can limit
the bandwidth of the analog data to be less than a certain frequency, 𝑓1, then the minimum
sample rate needed for reconstruction of this band-limited analog signal will be greater than
 270

twice the 𝑓1 value. For example, audio signals that range from 20-20,000 Hz need to be
sampled at a minimum rate slightly more than 40,000 samples per second. While this example
is correct, there is more details needed to make this work in practice.
The sampling theorem requires that the maximum frequency be limited to a value using
an electrical filter made of various components. From the theory of electrical filters, a perfect
filter that limits frequencies to an exact upper value cannot be built unless you have infinite
time to wait for the results. This means that real filters have rounded edges to the frequency
response and do not drop down to a zero value. There is a whole engineering discipline around
designing filters with various names used to describe their performance. The result from the
implementation of analog filter design is that you need a guard band or some extra frequency
range so that the filters performance is good enough for the analog sampling to work. When
the digital to analog conversion occurs at the end of DSP operations a filter is also needed to
limit the output frequencies. The filter on the end of the D to A conversion will keep unwanted
replicas of the audio frequencies at multiples of the sample rate from being produced. The
result of the need for practical electrical filters and to be at least higher than twice the highest
audio frequency is one of the reasons that CD music uses a 44.1 kHz sampling rate. Within
the audio community the subject of what is the needed rate for good music has prompted
many pseudo-science discussions and will not be discussed in this Handbook.
The science for sampling theory is solid and the principals developed by Shannon and
many others has provided many wonderful and useful products for us.

26.3 DSP Needs for Sample Rate

DSP filters are implemented using either an FIR (Finite Impulse Response) or IIR
(Infinite Impulse Response) topology. Each of these methods has advantages and
disadvantages but, in all cases, the digital operations are producing results at discrete values
determined by the sampling rate. Due to the properties of these filters the upper limit to the
filters response is the sampling rate divided by two. What this means is that if you have a DSP
low pass filter that stops frequencies at a value of 20,000 Hz and you are using periodic
sampling of 50,000 Hz then the filters performance for frequencies greater than 25 kHz will
not be a low pass at all. For DSP filters the response repeats at multiples of the sampling
frequency and these extra responses are limited only by the practical frequency response of
the electrical components in the hardware circuits. What this means for our restoration work
is that if we want to perform any filtering of the frequency content a high sampling frequency
will allow the DSP operations to keep working above the filters cutoff frequency and perform
 271

as expected. When a low pass filter is used to limit record noise above a value of 10,000 Hz,
we expect that this filter will continue to work to frequencies greater in value than 10,000 Hz.
So, what is a reasonable sampling value to use?
A practical sampling frequency for filtering work is to use the common 96 kHz value.
This sample rate will keep the filters working to at least 48 kHz. This high sampling rate is
not the minimum Nyquist sampling rate for the music; rather it is a result of using DSP to
implement filtering.
The various noise filters in the DCart10 software will also benefit from having many
samples to eliminate a noise event and replace it with music. 44.1 thousand samples per
second will work quite well but using 96 thousand per second seem to find and fix noise
events better than 44.1 thousand per second. The filters have more information to remove the
noise event and replace the noise with the correct music.

26.4 Caution with Sample Rate Selection in a Converter

While many converters will offer you the ability to select a sampling value of 96 kHz
this value will not always be the actual value that the device implements. To achieve a real
96 kHz rate, the needed clock signals and circuit frequency response have to operate at a high
speed. Some of the vendors have implemented a low-cost solution to high sample rates by
sampling at a lower rate and then adding extra digital values using special digital algorithms
to yield a pseudo 96 kHz rate.
Another problem area for accurately setting sample rate is the need for all the software
commands from the Diamond Cut DCart10 program to be implemented in the Converter used
in your Music Workstation. The chapter on transcriptions covers this subject in detail.
The chapter on Audio Measurements describes an experiment to verify that the
converter truly sampled the signal at the advertised rate.
 272

27.0 Audio Measurements

This chapter describes some useful measurements that you can perform using the
DCart10 software and laboratory equipment. These tests help evaluate the performance of
your audio system.

27.1 Measurement of Actual Converter Calibration

This experiment measured the relationship between the analog signals applied to the
Line-input and the recording level as shown by the DCart software. This experiment measures
the headspace and the nominal line-in value for the conversion. This information was needed
as there was no documentation available from the sound card vendor.
The sound card used Realtek components on the computer’s motherboard. For this test
an external 1 kHz sinewave was injected into the right channel of the line-input. The
amplitude of the sine wave was measured as the input level signal was increased from about
0.1 Volt P-P (Peak to Peak) to a maximum of about 1.0 Volt P-P. For this test the DCart10
software was running with the Record function on while set to a Pause state. In this setting
the VU level meters would indicate what the system was measuring as the amplitude was
changed. The measurements were in dB with all 1’s from the soundcard defined as 0.0 dB.
The sample rate was 44.1 kHz with a depth of 16 bits and was set by using the Microsoft
Control Panel’s Software along with the Record Tab and Advance Options. The results are in
table 27-1:

P-P Input dB Shown on Meter Calculated Calculated

dB(0.923) dB(0.894)
0.11 -20 -20.77 -19.17
0.23 -14 -13.83 -12.77
0.32 -10 -10.72 -9.90
0.42 -8 -8.16 -7.54
0.54 -6 -5.80 -5.35
0.62 -4 -4.50 -4.15
0.74 -3 -2.83 -2.62
0.82 -2 -1.87 -1.72
0.91 -1 -0.89 -0.82
1.05 0 0.46 0.42
Table 27-1
 273

A graph of the data showing calculated dB vs Indicated dB is shown in figure 27-1 for
the (0.894) reference case.

Input vs Output of A to D Converter

0.000

-5.000

O
u -10.000
t
p
u -15.000
t

-20.000

-25.000
-25.000 -20.000 -15.000 -10.000 -5.000 0.000 5.000

Input dB

Figure 27-1

A reference for the 0 dB value is required and two values were calculated. The 0.894
value comes from the assumption that the vendor used -10 dBv as a standard such that this
input would give all ones for an output. The value of 0.923 was found though experiment to
give the best fit to the actual data and is close to the 0.894 value.
Notice that the curve is not a straight line, as it should be. While the performance is not
perfect the result is close enough for music recording. From the test results, the Realtek
soundcard was designed to have a maximum output when the line-in signal had a value of -
10 dBv or 0.894 Volts P-P (Peak to Peak).

27.2 Verification of Sample Rate in Converters

The conversion of an Analog signal to a Digital value requites the input signal to be
sampled at a rate of so many samples per second. The sample rate of the Realtek converters’
range is from 11.025 kHz to 192 kHz for this device.
 274

The rate of conversion is directly related to the highest frequency that can both be
accurately converted to a digital signal and then converted back again to an analog signal.
The sampling of the signal must be at least greater than twice the highest frequency present
in the signal for accurate A to D and D to A conversion.
Some vendors will process the digital data after the conversion to add extra digital
values to simulate a high sampling rate. A test was developed to validate the actual rate vs
the indicated rate on the converter on the main computer module. The previous device
described in 27.1 was used.
The test consisted of applying a fixed amplitude sine wave signal to the line-in sound
card connector while the sample rate selected, and the frequency of the input sine wave was
varied. The DCart10 Record window was used in Pause mode. While the various signals were
applied the dB indicated level was monitored and recorded.
The idea behind the experiment was that if the actual sample rate was correct then the
level of the recorded signal should be steady until the frequency became close to or exceeded
twice the sample rate since the input level was held constant (-6dB on the meter) for all inputs.
The results are shown in Table 27-2 where the sample rate was the value selected for the
sound card and selected in DCart10 record window.

Sample Rate Sample Rate Sample Rate Sample Rate

Input Frequency 44.1kHz 48 kHz 96 kHz 192 kHz
1 kHz -6dB -6dB -6dB -6dB
10 kHz -6dB -6dB -6dB -6dB
15 kHz -6dB -6dB -6dB -6dB
20 kHz -9dB -7dB -6dB -6dB
25 kHz -40dB -16dB -6dB -6dB
30 kHz -6dB -6dB
40 kHz -6dB -6dB
45 kHz -8dB -6dB
50 kHz -12dB -6dB
70 kHz -6dB
95 kHz -6dB
100 kHz -6.5dB
105 kHz -8.0dB
Table 27-2
The results show that the sampling rate selected was the actual sample rate since the
output is constant to approximately one half of the sample rate.
 275

28.0 Conclusion and Reader Request

Writing this book has giving me an opportunity to share with you the joy that I have
received from restoring and then hearing old music sounding new again. The learning process
for me has been a long journey and I wanted to help you along on your journey too.
I have set the font size set to 14 so that you can read the information on the pages while
the book is on a table, as you have the computer running the Diamond Cut Productions
software. With the larger font size, I have used small margins on the pages to keep the number
of pages at a reasonable number. Also, I used a comb binding so that the book could be flat
on the table while in use. Please use any white space in the Handbook to record your notes.
I would ask that if you find errors in this Handbook to please send that information to
me via E-Mail. For that matter, let me know what you think about this book and if you would
like to have more information added to the book in the future.

My contact information is: hildebrantconsulting@outlook.com

Thank you for purchasing this Handbook!

Marc Hildebrant
Cape Cod, Massachusetts
USA