Audio Engineering

Associates

STUDIO RIBBON MICS

MIC POSITIONERS
TALL STANDS
ACCESSORIES

M-S Stereo: A Powerful Technique for

Working in Stereo*
By Wesley L. Dooley and Ronald D. Streicher
The practical requirements of broadcast and cinema stereo sound dictate the need
for good stereo imaging, as well as full monaural compatibility. Coincident miking
fullls this requirement, and the most versatile of these techniques in the M-S
matrixing of a forward-facing directional microphone with a laterally oriented
bi-directional microphone. The results offer both good stereo perspective and full
(discrete) monaural compatibility. The importance and implementation of this
technique to the recording, broadcast, and lm media are discussed.
INTRODUCTION
Since the earliest reported experiments in binaural sound reproduction, dating from the Paris Opera in
18811, the search has continued for improvements in stereo techniques.
Alan Dower Blumlein applied for a patent in 19312 that stands today as a benchmark in the history of
stereo technology:
The fundamental object of the invention is to provide a sound recording, reproducing and/or
transmission system whereby there is conveyed to the listener a realistic impression that
the intelligence is being communicated to him over two acoustic paths in the same manner
as he experiences in listening to everyday acoustic intercourse. This object also embraces
the idea of conveying to the listener a true directional impression and thus, in the case in
which the sound is associated with picture effects improving the illusion that the sound is
coming only from the artist or other sound source presented to the eye.
From 1931 to the present, the techniques developed by Blumlein and others have been expanded, rened,
rediscovered, and much discussed. This paper continues in that great tradition, with particular emphasis
on the applicability and appropriateness of the M-S technique to the recording, broadcast, and visual
media.
1 M-S STEREO: CONCEPT AND THEORY
The mid-side (M-S) stereo technique is one of the two formats of intensity stereo, that is, stereo in which
spatial localization is determined by the differences in the intensity of a sound wave as it arrives in phase
at a coincident pair of microphones. Intensity stereo relies completely on the directional characteristics
(polar patterns) of the microphone pair to produce this effect, since only intensity differences and not
phase differences exist between the channels for any single source arriving at a coincident pair.
1029 N. Allen Ave., Pasadena, CA 91104

www.wesdooley.com

Phone: (626) 798-9128 Fax: (626) 798-2378

The two format variations of intensity stereo are XY (left-right) and MS (mid-side or sum and difference).
XY stereo is created when a pair of directional microphones of like polare pattern are angled symetrically
from the midline (axis of symmetry) of the sound stage.
The MS technique likewise employs two coincident microphones: one is the mid (M) microphone, which
can be of any polar characteristicsfrom omnidirectional to cardioid to bidirectionaland which is aimed
directly at the midline (axis of symmetry) of the sound stage; the other is htre side (S) microphone, which
has a bidirectional pattern (cos ) that is oriented at right angles horizontally to the centerline axis of the
M microphone. The conversion of these two (midside or sum and difference) signals into convetional
leftright stereo is accomplished via a sum-and-difference matrix network, where typically the left signal
is the sum (L = M + S) and the right signal is the difference (R = M S).
In-phase intensity stereo provides the listener with a much more accurate and consistent stereo illusion
than do spaced microphone techniques. Intensity stereo provides angular position information that is
consistent, regardless of the distance of the sound source to the stereo microphone pair.
Spaced microphone techniques provide directional cures which vary radically with the distance of the
sound source from the microphones. When the sound source is relatively close to one of the microphones,
small changes in position result in dramatic stereo perspective shifts. When the sound source is distant
from the spaced micrephones, no directional information is conveyed.
Another disadvantage of the spaced microphone technique is that a sound source can arrive out of phase
at one microphone with respect to the other(s) when the air paths from the source to the microphones are
unequal. This is a major problem at lower frequencies where the soune source is at either the far left of
the far right of the stereo stage. When combined to mono either electrically or acoustically at the listening
poistion, considerable low-frequency comb ltering effects can occur. (These can be quite difcult to
detect when monitoring via stereo headphones, but become quite obvious via loudspeakers, or when the
signal is summed to mono.) During disk mastering this out-of-phase information translates into excessive
vertical motion of the cutterhead.
The choice of the polar pattern of the M component, and the relative proportions of the M and S contributions
to the whole sound eld determine the resultant stereo perspective. As the microphones are essentially
coincident with respect to the sound source, the stereo imaging is not phase dependent at any angle of
horizontal incidence; it is also independent of time (distance) related phenomena. Furthermore, when
microphones of suitably accurate polar frequency response are used, uniform response over the entire
sound eld is achieved.
Figs. 1 and 2 show some of the equivalent XY miking patterns and angles resulting from various
combinations of M-microphone polar patterns and M0toS ratios. These diagrams illustrate a number of
aspects of utilizing M-S stereo to derive XY stereo, as well as several more fundamental aspects of the
microphones polar patterns.1
The 81 diagrams illustrated in Fig. 2 are a representative selection of MS to XY transformations
In theory all basic microphone polar patterns can be described as a ratio of omnidirectional (scalar) to bi-directional (vector or cos 0) components
of two perfectly coincident microphones whose outputs have been summed. The polar pattern resultant can be expressed by the equation P
= B + (1 - B) cos), where p is the polar pattern resultant and B is valued between 0 and +1 and expresses the ratio of the omnidirectional to
bi-directional components. The typical resultant values and polar patterns for common rst-order microphones are shown in Table 1.

graphically presented, where X = M + S (the sum), and Y = M S (the difference). The diagrams are
organized into groups of nine, each group presenting a different M-microphone pattern of the MS
pair.
They are sequenced through nine ratios of M to S, from 0.30M:0.70S to 0.70M:0.30S, stepped in .05
increments. The relative size of the M and S polar plots in each MS pair graphically presents these ratios.
The equivalent XY patterns are shown below. The number of degrees printed between each XY pair is
the included angle between the axes of the microphone of that pair.
The equation given with each diagram describes that particular M-S pair by dening the ratio of M to
S and the polar pattern of the M microphone. The output of the S microphone is dened here as a sine
function (rather than the more conventional cos ) because its principal axis has been rotated 90 to the
left of the zero axis.
The general equations used to describe the M and S components are:
M = | A (B + (1 B) cos )

0 2

S = | (1 A) sin )

0 2

where A expresses the decimal fraction of the M microphones contribution to the MS matrix, and B
establishes the pattern of the M microphone by expressing the decimal fraction of the omnidirectional
(scalar) to bi-directional (vector) components of its polar pattern. Both A and B are dened as positive
numbers valued from 0 to + 1. Where A = 0, there is no M microphone contribution to the matrix; where
A = 1, the M microphone makes the entire contribution to the matrix. Where B = 0, the Mmicrophone
polar pattern is bi-directional (cos ); where B = 1, the Mmicrophone pattern is omnidirectional (1 or
scalar). Where B = 0.5, the M-microphone polar pattern is cardioid (0.5 + 0.5 cos ). (Note: the M and S
values were restricted to positive values for purposes of plotting the polar patterns. In fact, the values can
be negative, and these negative values represent the rear lobes (reversed polarity). The right facing lobe
of the S microphone, and rear-facing lobe ( rad) of the M microphone, and the rear-facing lobe of any
equivalent XY pair shown are all such reverse-polarity lobes.)
The equation used to dene the extremum (0 axis) of the X = M + S equivalent microphones angle
(one-half the included angle of the XY pair) is:

1 A

arctan
A
(
1

B
)

The derivation of this equation as well as the program run on an Apple II computer used to generate the
values given are available upon request from the authors.
The important aspect of an ideal omnidirectional (scalar) microphone is that it possesses no mechanism
for reporting the direction of a sound wave; it responds only to the (pressure) magnitude. The fundamental
aspect of a bi-directional (vector or cos 0) microphone patterns is that it reports both the magnitude
and polarity of a sound wave, based upon the angle of incidence: it has greatest in polarity magnitude
at 0 degrees (on-axis) incidence; equal magnitude but opposite polarity at 180 degrees incidence; and
minimum magnitude at both 90 degrees and 270 degrees incidence. When coincident omnidirectional
and bi-directional microphones are combined equally, the resultant is a cardioid.

Characteristic

Omnidirectional

Bidirectional

Cardioid

Hypercardioid

Supercardioid

Cos

1/2(1+Cos )

1/4(1+3Cos )

.37+.63Cos

360

90

131

105

115

360

120

180

141

156

-6

- 12

- 86

-6

-117

90

180

110

126

1
0 dB

333
- 48 dB

333
- 48 dB

250
- 60 dB

268
- 57 dB

17

17

19

Polar
Response
Patterns

Polar Equation
If () =
Pickup Arc
3 dB Down ()
Pickup Arc
6 dB Down
Relative Output
at 90 (dB)
Relative Output
at 180 (dB)
Angle at which
Output = 0 ()
Random Energy
Efcient (RE)
Distance Factor
(DSF)

Table 1. Polar equations a diagrams for rst-order microphone patterns [3].

In an M-S stereo pair comprised of an omnidirectional and a bi-directional microphone, the combination
will produce an equivalent X-Y pair of cardioid microphones oriented back to back (180 degrees) if the
M and S components of the sum-and-difference matrix are in a ratio of 50:50. From Fig. 1 (a) observe
that if the composite ratio favors the bi-directional microphone (the 30:70 ratio) results in a pair of backto-back subcardioids. Note that in all of these cases, the included angle between the resultant equivalent
X-Y microphone pair is always 180 degrees and only the polar patterns vary.
In the next situation, when the M microphone of the M-S pair is a cardioid (Fig. 1 (b)) and the ratio is
50:50, the resultant equivalent X-Y is a pair of hypercardioid microphones at an included angle of 126.9
degrees. As the M microphone becomes the dominant component (the 70:30 ratio), the included angle of
the equivalent X-Y pair becomes narrower (81.2 degrees), and the patterns become more nearly cardioid
(the rear lobes are reduced). When the S component is dominant (the 30:70 ratio), the included angle of
the equivalent X-Y pair broadens (155.8 degrees), and the patterns become more bi-directional (larger rear
lobes). Note that in these cases, both the included angle and the polar patterns of the resultant equivalent
X-Y pair are varied.
Finally, when the M microphone of the M-S pair is a bi-directional, the resultant equivalent X-Y patterns
are always bi-directional. Note here (Fig 1 (b)) that the only aspect that changes with the M-to-S ratio is
the included angle between the bi-directional pair. At the condition where the ratio of M to S is 50:50, the
included angle of the equivalent X-Y will be 90 degrees - the same as with the original M-S pair. The axis
of the polar patterns, however, is rotated 45 degrees with respect to the center axis of the sound source.
2 When the S component dominates (the 30:70 ratio), the included angle becomes quite large (133.6);
when the M component dominates (the 70:30 ratio), the angle narrows (46.4 degrees).

A complete set of cases showing the resultant equivalent X-Y pair resulting from a variety of M microphone
polar patterns and M-to-S ratios is diagrammed in Fig. 2. It can be seen that the M-S technique is a
transformation of the more conventional X-Y technique.
M-S stereo is not a panacea, but it does afford the diligent producer and/or mixer with the best set of options
available. Properly trimmed, the position of the sound source across the stereo stage can be accurately
recreated. Alternately, the stereo image can be altered to achieve a particular effect. The stereo image
can be compressed toward the center by increasing the M microphones contribution to the M-to-S ratio.
It can also be expanded toward the extremes by increasing the S microphones contribution. Finally, the
center can be expanded by broadening the M microphones polar pattern toward omni-directional.
2 THE PRACTICAL APPLICATION OF M-S
The use of the M-S technique offers many advantages over other miking techniques, for it provides a
discrete on-axis monaural pickup, together with a very versatile and potentially accurate stereophonic
image. Monaural compatibility is, by denition, absolute, since when the two stereo channels are summed,
only the signal from the M microphone remains: (M + S) + (M - S) = 2M. Thus the summed monaural
signal is the discrete output of the M microphone. Proper choice of the polar pattern and placement of the
M microphone will result in monaural quality that cannot be equaled by any other stereophonic technique.
By listening carefully to the M signal during microphone placement, this monaural compatibility can
be optimized. Another related benet is that with the M microphone aimed directly at the centerline of
the sound source and substantially on axis to it, the midsection of the source is not subjected to the offaxis coloration of image ambiguity common to many other stereo techniques (assuming of course, good
frequency and polar characteristics of the M microphone).
Similarly, since the reverberant sound eld is largely on axis to the S microphone, the coloration of the
reverberant eld is also reduced. In terms of the monaural/stereo compatibility, this has the pleasant result
that M-S stereo often provides for a smaller reverberation component in the summed monaural signal than
in the described left-right stereophonic signal. Since the ear is less tolerant of reverberation in a monaural
signal than in stereo, this reduction of the reverberant sound eld when summing to mono is benecial
for stereo radio, television, and lm production, where a large segment of the audience listens in mono.
Another advantage of the M-S system is that each component (M and S) is available to be treated
separately, prior to matrixing. This allows for selective correction of some problems encountered on
location such as out-of-phase low-frequency noise from the environment. Here the S component, which
contains the majority of this information, can be passed through a high-pass lter to reduce this unwanted
low-frequency component, and this can be accomplished without any alteration of the M component, as
long as the matrixed left-right image remains acceptable.
The exibility of the M-S technique is a great advantage to the sound engineer/producer both during
and after the recording session. In real-time situations, using a microphone with remote pattern selection
(such as the single-point stereo microphones available from AKG, Milab, Neumann, and others) the M
component pattern can be varied to optimize the pickup for any changes in the performing group or setup.
Similarly, given a xed setup, the M component pattern can be varied to optimize the pickup for any
changes in the performing group or setup. Similarly, given a xed setup, the M component pattern can be
varied to optimize the pickup for any changes in the performing group or setup. Similarly, given a xed
setup, the M component can be optimized by selection of a particular capsule with a xed pattern. Some
of the newer generation of smaller diameter capsules that possess a more uniform frequency response

off axis can be used. These improved polar pattern capsules (such as the Schoeps MK-41 hypercardioid,
which is widely used in lm and television productions where a tight pattern with good off-axis response
is desired) permit a more predictable congruence between the real and the ideal. Given a matrix network
with control over the ratio of M-to-S components in the output, the resultant stereo image width and
the reverberation component can also be manipulated. This ability to vary or select the pattern of the M
component and the image width allows the balancing engineer much latitude in control from the console
during real-time production.
Manipulation of the stereo image perspective can also be exercised during postproduction if the M and
S signals are recorded directly (as M and S) rather than via the matrix (as left and right). In this way, the
stereo image can be chosen or varied to suit the requisites of the moment (e.g., the visual perspective in
lm or television). Stereo/mono compatibility still remains completely predictable, because the discrete
monaural signal is directly available.
Sum-and-difference matrixing of the M-S signal to its equivalent X-Y outputs can be accomplished in
a variety of ways. The rst published method utilized two transformers with single primary and dual
secondary windings (a current example is shown in Fig. 3), where one of the secondary windings from each
of the transformers is connected in series and in polarity for the sum output. The other set of secondary
windings is connected in series but out of polarity for the difference output. If attenuators are inserted
prior to the transformers to adjust the incoming M and S signals, the ratio of M to S can then be varied
(Fig. 4). A difculty with this approach is the increased distortion and noise introduced by the added
transformers and pads in the signal path. (If constant-impedance pads are not used, the frequency and
phase response suffer from the varying load and source impedances introduced.)
An alternate method involves using three inputs on the mixing console; the M microphone is assigned to
both the left and the right channels equally, and the S microphone is split via a Y adapter to two inputs,
one assigned to the left channel and the other, with the polarity reversed, to the right channel. Thus the
left channel receives the sum of M and S, while the right channel receives the difference. By adjusting
the relative levels of M and +S signals, the matrixing of the signals is accomplished, with fewer of the
problems encountered with the microphone-level transformer matrix described above. The major problem
with this arrangement, however, is that the S microphone is now driving two loads and is seeing one-half
its normal terminating impedance. This can result in increased noise and/or distortion, depending on the
microphone and preamplier inputs involved. This is also an awkward technique to use when making
real-time decisions and varying M-to-S ratios.
Best results are achieved when the M and S microphones are brought directly to high-quality microphone
preampliers, and then are subsequently matrixed at line level. Again, this can be accomplished by
transformers and attenuators, but better results are achieved by using active components which provide a
well-balanced matrix and a means of control to vary the M-to-S ratio without compromising the signalto-noise, frequency, or phase response of the system. Console modules have been available in the past
from European manufacturers, such as Telefunken, E.M.T., and Calrec (5), (6). Most recently an outboard
active matrix control unit has become available in the United States (Fig. 5). This convenient self-powered
unit operates at line level, with a single-knob control to the M-to-S ratio (7).

3 PRACTICAL APPLICATION NOTES

The use of the M-S technique in no way replaces the need for
careful judgment in microphone placement. Proper balance
between the direct and the reected sound components of the
M signal must be made. When choosing the M microphone
pattern, the engineer confronts the dichotomic requirements
between monaural and stereophonic perspectives: the M
pickup can be too narrow for a good overall monaural
sound, but still provide an excellent stereophonic image
due to the contribution of the S pickup. Conversely, the M
pickup can also be too wide, allowing too much reverberant
information into the monaural signal. If the on-site engineer
pays particular attention to the aural quality of the M signal,
maximizing clarity and balance of direct-to-reected sound,
the derived stereophonic image will, in all likelihood, give
pleasing and reasonable accurate results. As stated earlier,
separate (discrete) recording the M and S signals will allow
for postproduction control of both of these signals, as well as
long-term archival retrieval and/or enhancement.
It should here be mentioned that the M microphone pattern
responds primarily to that part of the sound eld where the S microphone is most responsive (except
in the case where the M pattern is omnidirectional). Thus the M-S technique is ideal in the horizontal
plane. However, as the microphones are not absolutely coincident in the vertical plane, minor anomalies
do occur with respect to the vertical sound eld. As the ear is less capable of localization in the vertical
direction, however, and as these differences are minimal if the two microphone elements are kept very
close together, these anomalies are functionally insignicant.
Since the 0 degrees axis of the M microphone is aimed at the midpoint of the sound source, best frequency
response will be exhibited on axis, in the horizontal plane. This is another reason why the M-S technique
offers better monaural compatibility than its equivalent X-Y counterpart. The M microphone is primarily
on axis to the subject as a whole, as contrasted with the X-Y technique, where the principal axes of the
microphones are always off axis to the centereld of the subject, with the consequent coloration and
instability of center image inherent in may real-world microphones.
Directional microphones generally exhibit best frequency response when the sound source is directly on
axis and far enough away to avoid bass boost due to proximity effect. When a microphone is rotated so
that the sound arrives from progressively further off axis, high-frequency response typically decreases
correspondingly. Anomalies will also be observed which are caused by the acoustical effects of the
microphones all of these effects will be more pronounced with larger diameter capsule designs as compared
with smaller capsules.
Examining polar diagrams plotted over a range of frequencies for several widely used condenser
microphones (of both large and small capsule design) can be very enlightening. For a hypercardioid
pattern, the response at 45 degrees off axis is important, because a 90-degree X-Y pair of hypercardioid
microphones is a popular conguration. For a cardioid pattern, the response at 55 degrees off axis is of

interest, because a 110 degrees included angle pair is frequently congured with these microphones. The
off axis response at these angles represents the midline frequency response of the X-Y pairs so congured,
and illustrates why the center imaging and the monaural summation can be less than satisfactory, while
the sound from the extremes can be more distinct.
M-S stereo offers best frequency response at the center stage where the M microphone is on axis and the
extremes where the S microphone is on axis. As a solo pickup, the M-S technique is especially useful
not only because of the on-axis pickup of the M microphone, but also because the solo sound source is
located primarily in the null of the S microphone. The S component is therefore primarily reverberant
information, and thus the M-to-S ratio can be used as a direct-to-reverberant ratio control.
4 SUMMARY
The M-S technique, when used with a stereo microphone having remote pattern control capability and an
adjustable matrixing network, offers great exibility and versatility in stereo and possesses the additional
advantage of superior (discrete) monaural compatibility. This perspective selection is one of the virtues
of the M-S technique and can be of great advantage in both real-time and postproduction situations.
All miking involves some sets of compromises. Currently used as a recording technique for records, stereo
FM broadcast, and multichannel cinema productions (and in the future for stereo AM and television),
the M-S stereo technique requires fewer compromises to be made, and extends to the creative balancing
engineer and producer more useful options than any other stereo technique.
5 ACKNOWLEDGMENT
The authors with to express their gratitude to the following people who have contributed greatly to the
development of this paper: Sara Beebe for text editing; Kirby Fong for the expression of the M-S to X-Y
polar plots and equations; Deane Jensen for additional technical support and computer analyses; John
Kossey for photographic realizations of diagrams; Richard Knoppow for technical support and advice;
and Lois Lipton for additional graphic realizations. The authors also wish to acknowledge the support
and encouragement of the following people: Richard Burden, Marv Headrick, Richard Heyser, and Bill
Isenberg.
6 REFERENCES
(1) The Telephone at the Paris Opera, Sci. Am. (1881 Dec. 31)
(2) A.D. Blumlein, British Patent Specication 394, 325, 1931 Dec. 14; reprinted in J. Audio Eng. Soc.,
vol. 6, p. 91 (1958 Apr.).
(3) J. Eargle, Sound Recording. (Van Nostrand Reinhold, New York, 1976), pp. 53 ff., p. 125.
(4) D. Jensen, Data sheet for the JE-MB-D Transformer, Jensen Transformers/Reichenbach Engineering,
North Hollywood, CA.
(5) A. Nisbett, The Technique of the Sound Studio (Hastings House, 1962), pp. 68 ff., pp. 162 ff.
(6) J. Mosely, Eliminating the stereo seat, J. Audio Eng. Soc., vol. 8, p. 46 (1960 Jan.).
(7) Data Sheet for the MS-38 Active M-S Matrix Decoder, Audio Engineering Assoc., Pasadena, CA.
(8) H.A.M. Clark, G.F. Dutton, and P.B. Vanderlyn, The stereosonic Recording and Reproducing System
- A Two-channel System for Domestic Tape Recorders, Proc. IEEE (1957 Sept.); reprinted in J. Audio
Eng. Soc.,vol. 6,p. 102 (1958 Apr.).
(9) T. Faulkner, M-S: Another Purist Technique, Hi-Fi News and Record Rev., pp. 55 ff. (1980 Aug.).

7 BACKGROUND READING
(10) I.B. Weingartner, M-S Stereo Recording Techniques, H. Safransky, Trasl./Ed. (North American
Philips Corp.).
(11) Application Notes, db, p.47 (1980 Dec.).
(12) R. Streicher, Principles of M-S Recording, Audio Engineering Assoc., Pasadena, CA.
APPENDIX
For those who want to do additional reading in M-S stereo, the following references are recommended.
References (1), (2), (6)-(9) are of particular interest. (Note that (2) is the original paper in the eld and
proposes most of the relevant theory on the subject, (8) is an extensive recapitulation, and (9) is a very
concise summary.) The authors would be interested in obtaining copies of any oter related reference
materials or research work that may have been overlooked. If you have any such materials, please contact
us.

Other Products by Audio Engineering Associates:

Other Products by Audio Engineering Associates:
RCA Working Reproduction Microphones and replacement parts
AEA R44C Microphone
Our tribute to the classic RCA 44B using New Old Stock ribbon material
AEA 44CX Microphone
6db more output for critical digital recordings
AEA Ribbon Microphones
R84 - Studio Ribbon Mic
R88 - Stereo Ribbon Mic
R92 - Studio Ribbon Mic specically designed for close micing and guitar
Ribbon Mic Preamp
Modular Studio Microphone Stands and Booms
Since 1983 we have been the US agent for Coles Electroacoustics, manufacturers of the 4038 studio ribbon microphone and the 4104B, lip mic for voice-over work in high noise environments. We sell and
service the mics and stock replacement parts.
In North America we represent CB Electronics, a leading worldwide supplier of machine control equipment to the sound-for-picture industry. Their products specialize in professional control of and translation
between bi-phase, 9-pin serial and time code machines. Their SR line provides low cost multiple machine
remote controls for RS-422, Sony, and Tascam DA88 protocol machines.

Audio Engineering
Associates

STUDIO RIBBON MICS

MIC POSITIONERS
TALL STANDS
ACCESSORIES

1029 N. Allen Ave. Pasadena, CA 91104 Phone: (626) 798-9128 Fax: (626) 798-2378 www.ribbonmics.com