Noise Dose and Acoustic Shock from Headsets

Paul Darlington
Apple Dynamics
St Andrews Park, Queens Lane, Mold, CH7 1XB, UK

A paper prepared for …

Handset and Headset Testing – beyond narrowband,
114th AES Convention, Amsterdam, March 2003

Abstract

Headsets are capable of delivering wide bandwidth sound at amplitude that constitutes
a risk to hearing health. Whilst these devices have been in widespread use for many
decades as an audio entertainment source, recent changes in telecommunications and
IT have seen rapid change in the occupational use of headsets. These technological,
economic and cultural changes prompt re-examination of the risks of headset use,
particularly against the context of changes in the legislation that establishes
acceptable maxima for occupational noise exposure. It is the purpose of this paper to
describe the role of the headset in generating a noise exposure over prolonged periods
of use and in generating transient “shock” phenomena, both of which are known to
present health risks to the user. Means for practically instrumenting these effects are
discussed and some examples of real data are presented.

1 Introduction

Headsets, incorporating a sound source of circum-aural, supra-aural or in-the-ear type,

transduce an electrical voltage into a pressure signal audible to the wearer. Despite the
bandwidth limitations of telephone equipment, headsets are capable of generating
wide bandwidth sound of high quality and fidelity to the original voltage signal. This
fidelity implies a wide dynamic range. In use, the headset must generate sufficient
sound to overcome the masking effects of background noise (particularly in unilateral
and/or supra-aural situations). Given the wide dynamic range of the device, this places
amplitude peaks in the order of 0 dB Pa in ordinary use. Continued abuse of the
instrument, either by inappropriate or ill-advised selection of listening levels, or by
accident, makes the headset a potentially harmful noise source.

2 The Headset as Source of Occupational Noise Exposure

The greater number of descriptors of personal noise exposure are based upon the
“energy immission principle” by which the probability of long term hearing damage is
argued to be a correlate of daily acoustic energy “dose” (measured as an 8 hour A-
Weighted equivalent continuous sound pressure level). Under this principle, an
acceptable daily dose of Lep,8 = x might be established, which would allow a worker
to experience a higher level of x+3 for a 4 hour shift, a higher level still, of x+6 for a
2 hour shift, etc.. Metrics of personal noise exposure not based upon energy
immission (such as the OSHA recommendations) might associate the 3dB change
with other than a two-fold change in exposure time.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
2.1 Long-term Exposure

The noise dose delivered by a headset is determined by a number of factors. First of

these factors is the electro-acoustics of the headset itself. As with any loudspeaker, the
performance of a headset cannot be described simply in terms of a voltage to pressure
transfer function. The pressure generated by a headset when driven by an applied
voltage is unknown unless the load impedance is specified. In use, this load
impedance will be formed by the acoustics of the wearer’s outer ear and the fit of the
headset. The shape and size of individual’s ears is subject to wide variation [Ben],
with direct impact upon the load presented to a headset. The fit will also change from
one wearer to the next and will change over time with an individual. Given these
difficulties, it is more useful to measure the performance of headsets in the operating
conditions presented by standard ear simulators. (Omitting to model the individual
acoustics of a worker’s outer ear is standard practice in evaluating occupational noise
exposure).

Figure [1] shows the pressure sensitivity of a range of dynamic, supra-aural headset
receivers, measured on an IEC 318 artificial ear.

Figure 1 Sensitivities of a Range of Dynamic Headsets

The figure reveals the strong frequency selectivity of the headset types and a range of
absolute pressure sensitivities between (in this case) 17 and 30 dB re 1 Pa/V at 1000
Hz. Given the performance of a headset receiver, expressed as a magnitude transfer
function H(ω) of the sort reported in Figure [1], the noise exposure generated over an
interval T can be computed given knowledge of the statistics of the driving voltage.
As the headset and its load are frequency selective, it is necessary to calculate
frequency domain statistics of the voltage, such as the power spectral density Sxx(ω)
in order to compute the energy dose (which will be that experienced at the
microphone position of the artificial ear)….

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
  1 ∞


2π ∫ H ( ω ). S xx ( ω ) d ω 
L EP , d , ear = 10 log  −∞ 
 2
p ref 
 
 

in which pref is the acoustic reference pressure and the driving voltage power spectral
density is computed over an 8 hour average.

In order to relate the dose (above) to legislative limits associated with occupational
noise exposure, it is necessary to calculate the equivalent free-field pressure (rather
than that at the ear drum or at the microphone of an ear simulator). This is achieved
by a further frequency domain factor G(ω), such as that defined in ITU-T P-58 [1] or
otherwise, as appropriate. This yields:

 1 ∞


2π ∫ H ( ω ) G ( ω ). S xx ( ω ) d ω 
L EP , d , free − field = 10 log  −∞ 
 2
p ref 
 
 

These measures of personal exposure may be calculated by estimating the Power

Spectrum and performing the integration “off-line”, after the spectrum average is
completed. Alternatively, the effects of H and G may be realised as digital filters and
the computation performed on-line, allowing a running estimate of the instantaneous
equivalent free-field pressure to be computed from the driving voltage, which may be
averaged in real time. These methods are contrasted in [2].

2.2 Transient Effects; Peak Exposure Limits

Although the 3 dB / factor 2 in time trade-off above might suggest that there is no
limit to the safe noise exposure, provided that it is short enough in duration, there is a
peak pressure exposure limit. In the United Kingdom this limit is set to 200 Pa (46 dB
Pa). Whilst it is difficult to limit noise exposure due to long-term use of headsets, it is
possible to fit devices such as “acoustic shock suppressors” [3], which seek to limit
the peak pressure.

ITU-T P.360 [4] establishes limits on the pressure that headsets can generate at the ear
reference position of standard artificial ears. This standard is based directly upon a
LEP,d = 85 dB(A) limit, which is then corrected to account for several factors. Firstly,
the limit is reduced by 10 dB since “the acceptable noise level in the work place is not
applicable to non-occupational exposure”. Secondly, headsets are penalised with
respect to handsets by 7 dB, since headsets are (assumed to be) used for longer
periods. Thirdly, the limit is reduced by 4 dB to account for the band-limited nature of
the telephone signal. Finally, the limit is corrected by 5 dB to account for the
difference between ERP and free field measurements.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
The limits defined in ITU-T P.360 are calculated for both “longer duration
disturbances” (of 2 second duration) and “short duration impulses” (of 80 msec
duration, where the A-Weighting is no longer relevant):

“Longer duration disturbances” “Short duration impulses”

Headset 24 dBPa(A) 39 dBPa

Shock suppression is usually achieved on a headset using simple measures such as a

pair of shunting diodes, which limit the voltage developed at the receiver terminals. If
shock suppressors are designed and working correctly, it is impossible for a headset to
generate pressures that exceed the peak limit (shock suppressors are usually set to
operate at around 26 dB Pa, some 20 dB below the peak limit. Under these conditions,
the headset does not pose a risk to hearing health as understood in terms of transient
noise exposure. There are, however, repeated cases where exposure to transient
sounds in telecommunications apparatus has been cited as a causal factor in ill health
or chronic disability.

2.3 Acoustic Shock

Whilst a correctly operating shock suppressor should prevent signals that contravene
the peak action levels of noise at work regulations, telephone users are reporting
health problems associated with transient signals in phone systems. ITU-T P-10 [3]
defines “Acoustic Shock” as “Any temporary or permanent disturbance of the
functioning of the ear, or of the nervous system, which may be caused to the user of a
telephone earphone by a sudden sharp rise in the acoustic pressure produced by it”.
Other workers [5] prefer to speak of Acoustic Shock Injury (“ASI”) as naming a
group of symptoms, typically seen amongst telephone call centre workers, which arise
after exposure to transient acoustic signals - these signals are then called Acoustic
Shocks (this nomenclature is adopted in the present paper). The continuing disabilities
claimed after ASI include disturbance of auditory function, such as threshold shift and
tinnitus. However, the list also includes a wide range of psychological, behavioural
and social changes.

3 Examples of Real Noise Exposures

Gathering evidence of noise exposure associated with headset use is time consuming
and expensive. Some examples are presented below demonstrating a range of data
gathering methods; a large survey, sample data sets from beta testing of a new
monitor instrument and data from control centres’ own logging recorders.

3.1 Long-term Noise Dose

Two classes of data collected from headset wearers working in call- or control-centres
are presented. The first data set is the summary of a major study of UK call centre
workers [6], in which 150 workers were monitored. The survey was conducted using
a measurement technique in which a second test headset was driven by the same
voltage as that driving the worker’s headset. The second headset was worn by a
HATS (thus realising he headset transfer function H(ω) directly). The pressure
detected by the microphone in the artificial ear was averaged, in third octave bands
and the free-field correction G(ω) and A Weighting applied “off-line”.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
The workers surveyed were found to select listening levels on their headsets between
65 and 88 dB(A) with a mean of 77 dB(A). The distribution is shown in Figure 2.

Figure 2 Listening Levels in UK Call Centre survey

Given the pattern of work, the mean Daily Personal Noise Exposure in each call
centre ranged between 68 and 84 dB(A) with a mean of 75 dB(A). 3 of the 150
workers surveyed had a dose exceeding 85 dB(A).

The second data set is initial results from tests of a new monitor instrument [7], which
monitors the voltage driving the headset and uses Digital Signal Processing
techniques to derive a running estimate of the equivalent free-field pressure, from
which a dose is computed. The headset transfer function H(ω), the appropriate free-
field correction G(ω) and the A Weighting network were all applied in real-time and
the acoustic doses recorded every minute.

Two Fire Service Control Room operators were monitored. During their shifts they
handled both telephone and radio traffic on their headsets. Figures 3 & 4 below show
segments of two operators’ shifts. Operator 1 has chosen to use a higher listening
level (an in-line volume control is available), although the pattern of work includes
more breaks. Operator 2 has selected a lower listening level, but has worked a more
intensive shift. The equivalent continuous levels for the durations instrumented are
81.1 and 79.0 dB, respectively.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
Figure 3 Equivalent Free-Field Pressure, Operator 1

Figure 4 Equivalent Free-field Pressure, Operator 2

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
3.2 Acoustic Shock

Transient signals are, by definition, short lasting. Those transient signals caused by
rare fault conditions on a telephone system will, by definition, be difficult to
instrument, as it is unlikely for instrumentation to be connected to the appropriate line
at the moment the event occurs.

Fortunately, many call-centres and service-centres record a fraction of their telephone

traffic for e.g. training and monitoring purposes. In emergency call centres all traffic
is recorded. This gives access to data sets that include “Acoustic Shocks”.
Unfortunately, the data is severely compressed before storage, such that accurate
reconstruction of the original waveform is compromised. Further, the recording level
is never calibrated, such that reconstruction of the original amplitude is impossible.
With these limitations in mind, two examples of transient signals that have been
called “Acoustic Shocks” (by the operators who experienced them) are discussed
below.

The signals were recorded in an emergency service control room [8]. Both records
consist of an unexpected tone burst (the shock event) followed by a signal that is
useful in establishing the relative amplitude of the shock event.

The first record begins with a tone burst of some 125 msec duration, centred at
approximately 3.75 kHz, followed by a series of 11 DTMF tones. The tone burst is
approximately 3 dB above the amplitude of the DTMF tones.

The second record begins with a tone burst, again of approximately 125 msec
duration, but centred at approximately 3 kHz. The burst is followed by speech
(complaining of the sound – “Ow – did you hear that?”) against a dial tone. The tone
burst is 14 dB above the long-term speech level. The greater part of the energy in the
speech signal is contained in the 250 Hz octave band. If the listening level was set
with reference to such a signal, the subjective loudness of the tone burst would be
greater than that associated with a +14 dB increase in level, due to the frequency
dependence of the loudness percept.

4 Conclusions

It has been shown that a headset, of the sort used in telephone call-centres, service-
centres and control rooms, is capable of generating high acoustic pressures at the
eardrum. If these sounds are experienced over long periods they constitute a noise
dose, which may present a risk to the wearer’s long-term auditory health. Whilst
capable of delivering potentially dangerous energies over extended periods, the
headset should be incapable of generating dangerous transient pressures if fitted with
correctly functioning “acoustic shock suppressors”.

Those users complaining of injury resulting from “Acoustic Shocks” are unlikely to
have been injured by the energy in the transient signals they have reported. They may
have suffered injury from longer-term noise exposure or their injury may have been
caused by a process not to be understood in terms of conventional models of noise
induced hearing loss.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002
References

[1] ITU-T Recommendation P58 Head and torso simulator for telephonometry.
(08.96)

[2] Darlington P Practical Measurement of Telecommunication Receiver

Electro-Acoustics for the Computation of Acoustic Dose To be presented at: Call
Centres – a Measurement Headache: a one day meeting organised by the
Measurement & Instrumentation Group, Institute of Acoustics, Liverpool, UK, 5 June
2003

[3] ITU-T Recommendation P.10 Vocabulary of terms on telephone

transmission quality and telephone sets. (03.93)

[4] ITU-T Recommendation P.360 Efficiency of devices for preventing the

occurrence of excessive acoustic pressure by telephone receivers (12/98)

[5] Milhinch J Risking Acoustic Shock Risking Acoustic Shock Conference,

Fremantle, Western Australia, September 2001

[6] Broughton K A Measuring the Noise Exposure of Call Centre Operators

Risking Acoustic Shock Conference, Fremantle, Western Australia, September 2001

[7] Tyler R G A new non-invasive monitor to measure noise exposure from

headsets. To be presented at: Call Centres – a Measurement Headache: a one day
meeting organised by the Measurement & Instrumentation Group, Institute of
Acoustics, Liverpool, UK, 5 June 2003

[8] Acoustic Shock records provided by Keith Broughton, UK Health and Safety
Executive, used with permission.

Noise Dose and Acoustic Shock from Headsets Paul Darlington, www.appledynamics.com 2002