Audiometric Calibration: Speech Signals

Graham Frost, B.Sc., M.Sc. and Harry Levitt, Ph.D.

ABSTRACT

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Speech audiometry is used primarily for determining an
individual’s receptive auditory communication abilities. Speech also is
used for site-of-lesion testing. The speech signal is temporally and
spectrally complex, and there are unique challenges in acoustically
characterizing speech signals. This article describes some of these
challenges and details the international standards developed to specify
the requirements for these characterizations and to perform reliable test
protocols and measurements.

KEYWORDS: Speech audiometry, speech material, speech

detection threshold, speech level, long-term average speech
spectrum

Learning Outcomes: As a result of this activity, the participant will be able to (1) describe the history of
speech testing and the requirements as specified in current relevant standards; (2) assess the integrity of
speech test material presented to a test person; (3) list the issues that may affect the reliability of responses
to speech test material.

M ost audiometric test signals and stimuli can provide additional information with regard
are used to either confirm a hearing loss or to to an individual’s ability to detect, discriminate
gain information regarding the nature, severity and understand speech. These tests can also be
and possible cause of the hearing loss (differen- used in the evaluation of nonorganic hearing
tial diagnosis).1,2 These tests alone may not give loss.
a reliable indication of a person’s ability to A speech test consists of three major com-
understand speech. Two individuals can pro- ponents: the speech material (e.g., syllables,
vide similar pure tone audiograms and aural words, sentences, or conversational speech),
acoustic impedance/admittance test results, yet the method of signal delivery (by audition
their ability to understand speech may differ and/or vision) and the experimental protocol
significantly. The use of speech tests, however, (e.g., open or closed response sets, rating

1
Technical Consultant, United Kingdom; 2Department of Standards and Calibration. Part 1: Standards Process,
Research, Advanced Hearing Concepts, Bodega Bay, Physical Principles, Pure Tone and Speech Audiometry;
California. Guest Editor, Robert Burkard, Ph.D.
Address for correspondence: Graham Frost, B.Sc., Semin Hear 2014;35:346–360. Copyright # 2014 by
M.Sc., 131 Lower Church Road, Burgess Hill, West Thieme Medical Publishers, Inc., 333 Seventh Avenue,
Sussex, RHI5 9AA, United Kingdom New York, NY 10001, USA. Tel: +1(212) 584-4662.
(e-mail: [email protected]). DOI: http://dx.doi.org/10.1055/s-0034-1390163.
ISSN 0734-0451.
346
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 347

procedures and methods of scoring). Speech communicate over long distances led to the
materials that use appropriate methods of signal invention of the telephone. The telephone
delivery also are used extensively in patient addressed this need successfully and telephone
rehabilitation, including auditory training, networks soon expanded rapidly. The rapid
hearing therapy, and hearing aid fitting. growth of telephonic communication created
The focus of this article is on the delivery of a vast range of new necessities. These included
audio speech signals. methods for evaluating and standardizing tele-
Video speech communication using mobile phones, telephone lines, telephone exchanges,
phones or computer links presents a new chal- and the many other components of a telephone
lenge in establishing standards for this rapidly system. In addition, as telephone systems grew

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
growing form of communication. The stan- in size and complexity, new technical problems
dardization of video speech signals is beyond were encountered, such as poor speech intelli-
the scope of this article. At present, there are no gibility over very long telephone lines, necessi-
audiometric standards for video speech signals. tating the invention of new methods and
The final section on future developments dis- technologies to address these problems. The
cusses some of the issues that need to be focus of this article is on the standardization of
addressed in developing video speech standards. audiometric speech signals. It is thus appropri-
This article will review current published stan- ate to begin with a brief history of speech
dardization relating to the characterization and audiometry and how each advance addressing
recording of audiometric test material and the an important need begat new necessities, re-
methods of presentation of this material to a sulting in ever increasing needs for new and
test person. This article will also examine improved standards. There is some overlap with
current methods of monitoring, measuring, other articles in this issue, such as the standard-
and recording a test person’s response. ization of couplers and ear simulators used in
The calibration requirements for speech the calibration of earphones. In the case of an
test material will also be described together overlapping area of interest, this article provides
with how speech test protocols may be validated a brief overview of the issues involved from the
and, where appropriate, the acquisition of nor- perspective of speech audiometry, allowing the
mative data. In practice, live-voice presentation reader to gain an in-depth understanding of
of a speech test is often used, but this practice is how the issues are addressed from the primary
not recommended because of the very high article dealing with the topic.
variability of spoken test material. There are The rapid growth of telephonic communi-
significant test–retest differences in the tempo- cation created the need to evaluate speech
ral and spectral characteristics and level of signals in the telephone network and establish
spoken test material as well as substantial differ- standards for their measurement and calibra-
ences between testers such as male or female tion. The speech tests developed to address this
voice, accent, and dialect. These factors greatly need were found to be very useful in other areas,
influence the test and retest reliability and such as assessing the effect of hearing loss on
repeatability. Live voice is therefore not ad- speech understanding. In addressing this need,
dressed by current standards and should not be modern speech audiometry was born.
used for diagnostic purposes. The first speech test for evaluating speech
The limitations of current standardization communication over a telephone circuit was
will be discussed and areas requiring additional developed by Campbell in the early 20th cen-
investigation and documentation identified. tury.3 Nonsense syllables were delivered by live
voice using a telephone mouthpiece and a
telephone earpiece at the receiving end. These
HISTORICAL BACKGROUND: early tests were referred to as articulation tests,
TELEPHONES AND THE BIRTH OF although they were used primarily to evaluate
MODERN SPEECH AUDIOMETRY speech reception rather than speech production.
Necessity is the mother of invention and inven- Harvey Fletcher and the research group he
tion fathers new necessities. The human need to headed at the Bell Telephone Laboratories
348 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

dominated the early development of speech day communication, such as sentence-length

audiometry.4,5 His pioneering book, Speech material and continuous speech. A recent ad-
and Hearing,4 and the classic article with Stein- vance has been the development of adaptive
berg6 provided the scientific basis for the tests under computer control and auditory-
emerging field of speech audiometry. Although visual speech tests. For more information on
Fletcher’s research was driven by the need to the early history of speech audiometry, see
improve telephonic communication for people Levitt.10
with normal hearing, he recognized that his To administer speech tests reliably in the
research also would be of value to people with evaluation of telephone circuits, standards were
hearing loss. In the article with Steinberg, developed for the conversion of the acoustic

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
Fletcher and his coauthor point out that the speech signal, as produced by the human
test materials they had developed “were useful mouth, to electrical signals for transmission
not only in determining defects in transmission, over a telephone line. Similarly, standards also
but defects in the production and reception of were developed for the conversion of electrical
speech as well.”6(p.806) Speech audiometry thus signals to acoustic signals delivered to the
began as an offshoot of speech tests and mate- human ear. These standards necessitated the
rials developed for people with normal hearing. development of a standardized artificial mouth
The next major advance in speech audiom- and a standardized artificial ear (now known as
etry took place in the early 1940s when the either an acoustic coupler or an ear simulator,
Psycho-Acoustic Laboratory (PAL) at Harvard depending on how accurately its acoustic im-
University worked on methods for evaluating pedance approximates that of the human ear).
military speech communication systems. The The artificial mouth plays an important role
speech materials developed for this application in developing standards for recording speech. Of
were subsequently modified for use in speech particular importance for audiologists is the use
audiometry and standardized recordings were of an acoustic coupler or an ear simulator for
made of these test materials.7,8 These were the calibrating earphones in specifying sound levels
first standardized recordings to be developed in delivered to the human ear. For many years, an
speech audiometry. The Central Institute for acoustic coupler with an invariant acoustic im-
the Deaf (CID) W22 monosyllabic word test, pedance roughly equal to that of the human ear
as well as several other widely used audiometric was used for earphone calibration.11
speech tests, emerged from the research initiat- The acoustic impedance of the ear, however,
ed at PAL. Here again, an important advance in varies with frequency, and significant errors are
speech audiometry resulted from tests initially likely when calibrating earphones at high fre-
developed for people with normal hearing. quencies using an invariant acoustic coupler.
Carhart pioneered the development of Zwislocki developed an ear simulator that mimics
speech tests designed specifically for audiologi- the changes in acoustic impedance of the human
cal applications.9 He introduced the idea of ear with frequency.12,13 The American National
speech testing in noise because listening to Standards Institute (ANSI) and International
speech in noise is commonly encountered in Electrotechnical Commission (IEC) have devel-
everyday life and is particularly challenging to oped several standards for acoustic couplers and
people with hearing loss. He also observed in ear simulators depending on the type of trans-
his experimental studies that tests of speech in ducer that is used and the relative accuracy of the
noise were more sensitive to differences among measured sound levels. See Dillon for a discussion
hearing aids and also more sensitive to individ- of the various couplers and ear simulators that
ual differences in speech understanding in have been developed and relevant ANSI and IEC
people with hearing loss. standards.14 See also Zwislocki for an illustrative
In subsequent years, speech audiometry account of the development of the Zwislocki ear
grew in scope with the development of diag- simulator (also known imprecisely as the Zwislocki
nostic tests. There was also a trend away from coupler).15 The Zwislocki ear simulator is sold
the use of syllables and single words and a commercially as the Knowles DB100 ear simula-
greater focus on speech representative of every- tor. Other commercially available ear simulators
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 349

are the Bruel & Kjaer 4157 (Brüel & Kjaer Sound used as a screening test for identifying young
& Vibration Measurement A/S, Naerum, children within the general school population
Denmark) and the GRAS RA 0045 (G.R.A.S. who may have a hearing problem. Screening in
Sound & Vibration A/S, Holte, Denmark). this form had been introduced in the late 1930s
For detailed information on standard and was regularly used by educational authori-
methods of calibrating earphones and related ties until around 1950. Performing speech
transducers, see the article in this issue by testing cannot, in general, provide the same
Fedtke and Grason. The recent development information regarding the nature of a hearing
of powerful new methods for determining the loss that can be derived from other audiometric
acoustic impedance of the ear by means of test protocols.

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
reflectance measurements is likely to affect Everyday speech contains several clues,
the future development of standards for mea- including contextual cues, which enable the
suring and specifying sound levels in the ear (see listener to identify words and sentences even
the article by Wilber, Laukli, and Burkard in without being able to hear or discriminate each
this issue). Speech is a complex signal with individual component.
complex fluctuations in both its temporal and Speech test materials can be designed with
spectral structure. This presented difficulties in particular emphasis on the inclusion of nonre-
measuring speech levels conveniently and reli- dundant components. More reliable results
ably prior to the introduction of digital tech- with regard to assessing an individual’s true
nology. The volume-unit (VU) meter was an ability to discriminate can then be obtained by
analog instrument developed for the specific removing, where possible, clues that may help
purpose of measuring speech levels16 the test person identify a word or phrase. An
The ballistics of the meter were designed to extreme example of this is the use of nonsense
track major temporal fluctuations of speech at a words or phrases.
rate that can be followed by the human eye. It was There is now a diversity of speech test
a useful instrument, but it had limitations in that materials available, each designed for a specific
there were large between-operator differences in application and to provide specific information
reading the meter. There were also inaccuracies in that can be used in the assessment of an
measuring speech levels in the presence of back- individual’s hearing ability. The design of
ground noise. The introduction of digital tech- speech tests is not only the subject of continuous
nology allowed for the development of meters research but also continuous criticism. Unlike
that could measure, record, and analyze speech other audiometric test signals and stimuli,
signals with considerable flexibility. Improve- which are in general periodic narrowband or
ments included the capability of measuring true of short duration and are relatively straightfor-
root-mean-square (rms) values, instantaneous ward to characterize, speech signals are complex
peak values, and other aspects of the speech signal and broadband and have temporal changes in
that presented difficulties for analog instrumen- their spectra and intensities.
tation. Response times could be adjusted dynam- The results obtained from speech audiom-
ically and stored efficiently for subsequent etry depend upon the speech material and the
analysis. The effect of background noise and other test method used; this includes the method of
nonspeech signals on the measurements could be presentation to the test subject and the method
estimated and compensated for so as to obtain a by which a response is measured. Speech test
more accurate measurement of speech level. material is available in different formats, each
These advances rendered the VU meter obsolete designed to provide the audiologist with spe-
with its complicated analog circuitry.17 cific information with regard to the test person’s
hearing impairment and his or her ability to
detect and discriminate speech. Speech test
INTRODUCTION TO SPEECH material, normally referred to as speech material,
AUDIOMETRY consists of an entire set of test items, which can
Prior to the development of standardized be monosyllabic or polysyllabic words, logatoms
speech tests, testing with speech materials was (nonsense syllables), sentences, or timed
350 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

segments of connected speech. A response may (2012), part of the ISO 8253 series of standards
be recorded or scored on a complete test item or on audiometric procedures. (Note: A list of all
a specific part or parts of it, for example, standards relevant to speech audiometry, and
syllables within a word. The speech material referenced in this article, can be found
test items also may be open set, where the in Table 1.) At this time there are no published
number of alternative responses to each test ANSI Standards that cover this topic.
item is unlimited, or closed set, where the IEC 60645-2 (1993) specifies the general
number of alternative responses to each test requirements for speech audiometers. This
item is limited. standard makes provision for the use of both
Speech material can be delivered to the test recorded and live voice, and although it recog-

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
person by means of monotic, diotic, or dichotic nizes that a live-voice method of speech mate-
modes of sound delivery using various types of rial presentation may be frequently used, it is
transducer, such as supra-aural or circumaural clear that live-voice speech audiometry may not
earphones, insert earphones, or loudspeaker. be able to comply with the requirements of the
Monotic sound delivery uses a single transducer standard. A specification is therefore included
on one ear. Diotic sound delivery uses two to ensure a high degree of reliability. IEC
transducers, one on each ear with identical 60645-2 (1993) does not, however, include
signals at each ear. In dichotic sound delivery, requirements for the test environment. This
the signals at each ear are not identical. In standard also recognizes the fact that although
addition to speech being delivered by means of bone vibrators may be used to present the
acoustic transducers, bone-conduction trans- speech material to a test subject, their perfor-
ducers are used for diagnostic testing and for mance when using speech signals can be ex-
prosthetic aids utilizing bone conduction, such tremely variable and are not considered
as the bone-anchored hearing aid. Audiometric appropriate for standardization.
speech signals are also delivered by means of IEC 60645-2 (1993) does not include the
electrical stimulation in cochlear prostheses. requirements for systems that may be used to
Each of these modes of presentation has relay voice for test methods that require a
unique calibration issues, and measures must be spoken response from the test subject and
taken to ensure that the integrity of the signal where the tester is not in the same room as
delivered to the test subject is maintained. the subject. However, the performance char-
Speech audiometry also may be performed in acteristics of a voice relay system can signifi-
conjunction with the presentation of competing cantly affect the reliability of test results. In
noise to reproduce a more typically normal addition to specifying the general performance
listening environment. requirements of speech audiometers, IEC
In addition to the wide range of speech 60645-2 (1993) also specifies the requirements
material available, a test person’s response also for testing and calibration. The standard lists
may be measured or scored in several different the minimum features required for speech
ways. These include the test person repeating audiometers, which are classed as type A or
what they have actually heard and forced-choice type B depending on their functionality. There
responses, using several displayed words or are also additional type designations type A-E
images. Each of these methods presents its and type B-E where the audiometer also pro-
own issues with regard to the reliability of the vides a free-field equivalent earphone output
response and its use in making an assessment of where the sound pressure level (SPL) produced
an individual’s hearing ability. by the earphone in an acoustic coupler or ear
The international standardization of issues simulator is corrected by the difference be-
relating to speech audiometry are in general tween the close-coupled and the free-field
addressed by IEC 60645-2 (1993), part of the sensitivity levels of the earphone.
IEC 60645 series of standards specifying the For type A and type B audiometers, where
performance requirements of a range of audio- free-field equivalent earphone outputs are not
metric equipment, and the International Orga- provided, the output SPL and frequency re-
nization for Standardization (ISO) 8253-3 sponse using the earphone as measured in a
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 351

Table 1 IEC, ISO, and ANSI Standards Relevant to Speech Audiometry

ANSI C63.19 2011 American National Standard Methods of Measurement of Compatibility

between Wireless Communications Devices and Hearing Aids
ANSI S3.5 1997 (R2012) American National Standards Method for Calculation of Speech Intelligibility
Index
ANSI S3.6 2010 American National Standards Specification for Audiometers
ANSI S3.7 1995 (R2008) American National Standards Method for Coupler Calibration of Earphones
ANSI S3.25 2009 American National Standards Specification for an Occluded Ear Simulator
ANSI S3.55 2014 American National Standards Ear Simulator for the Measurement of Supra-aural
and Circumaural Earphones

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
IEC 60118-13 2011 Electroacoustics—Hearing aids—Part 13: Electromagnetic compatibility (EMC)
IEC 60318-1 2009 Electroacoustics—Simulators of human head and ear—Part 1: Ear simulator for
the measurement of supra-aural and circumaural earphones
IEC 60318-3 1998 Electroacoustics—Simulators of human head and ear—Part 3: Acoustic coupler
for the caliberation of supra-aural earphones used in audiometry
IEC 60318-4 2010 Electroacoustics—Simulators of human head and ear—Part 4: Occluded-ear
simulator for the measurement of earphones coupled to the ear by means of
earinserts
IEC 60318-5 2006 Electroacoustics—Simulators of human head and ear—Part 5: 2cc3 coupler for
the measurement of hearing aids and earphones coupled to the ear by means
of ear inserts
IEC 60645-1 2012 Audiological equipment—Part 1: Equipment for pure-tone audiometry
IEC 60645-2 1993 Audiometers—Part 2: Equipment for speech audiometry
IEC 61260-1 1995 Electroacoustics—Octave-band and fractional octave-band filters—Part 1:
Specifications
IEC 61672-1 2002 Electroacoustics—Sound level meters—Part 1: Specifications
ISO 266 1997 Acoustics—Preferred frequencies
ISO 8253-3 2012 Audiometric test methods—Part 3: Speech audiometry

Abbreviations: ANSI, American National Standards Institute; IEC, International Electrotechnical Commission; ISO,
International Organization of Standards.

National and International Standards are subject to regular review and revision. The requirements that are specified
in this article have been taken from current editions and are subject to change. It has recently been agreed that IEC
60645-2 (1993) will be merged with IEC 60645-1 (2012) to produce a new edition of IEC 60645-1, which will specify
performance requirements for both pure tone and speech audiometers. When the new edition is published, and as
other referenced standards are revised, some of the information contained in this article may no longer be applicable.

specific acoustic coupler or ear simulator (the Speech material recordings are required to
uncorrected earphone output level) must be carry a calibration signal and when the calibra-
specified by the manufacturer. The output tion input signal is adjusted such that the
SPL and frequency response of the audiometer indicated level is at the reference-level marker,
when used with a loudspeaker also should be the resulting SPL from the output device
specified, but as speech audiometers are often should not be more than
3 dB from the level
used with an external amplifier and a loud- specified for that particular speech material.
speaker specified for the test environment, these IEC 60645-2 (1993) specifies the permitted
characteristics may need to be derived during tolerances when measuring the output SPL of a
calibration. Speech audiometers also are re- loudspeaker for the purposes of measuring the
quired to provide a signal-level indicator that frequency response. Within the range 250 to
incorporates a reference level marker that can be 4,000 Hz, the SPL generated by a loudspeaker,
used during calibration; this is normally pro- when measured using a one-third octave filter
vided by a meter or bar display. This reference and a test signal at the same center frequency,
level marker may be used in the monitoring of a should not deviate from the average SPL
live-voice input. of frequencies within this range by more than
352 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

5 dB. For test signals from 4,000 Hz to sition and validation of speech test material.
6,300 Hz, the permitted maximum deviation ISO 8253-3 (2012) specifies a method for
is
8 dB. determining the speech detection threshold
The tolerances permitted when measuring level, which for a given test subject, for specified
the SPLs generated by earphones, however, are speech material, and for a specified manner of
specified in IEC 60645-1 (2012). The output signal presentation is the level at which speech
SPL produced by the earphone when measured is detected in 50% of the trials. It also specifies a
in the same acoustic coupler or ear simulator method for determining the speech recognition
used for its overall calibration, and using a one- threshold level that, for a given test subject, for
third octave filter and a test signal at the same specified speech material, and for a specified

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
center frequency, is required to be within the manner of presentation, is the lowest speech
range of 250 to 4,000 Hz and cannot deviate level or speech-to-noise ratio at which the
from the average SPL of all the test frequencies speech recognition score is 50%.
within that range by more than
4 dB. For
frequencies above 4,000 Hz, the permitted
maximum deviation is þ4 dB and 6 dB. CALIBRATION CONSIDERATIONS
These conditions are required when using all When considering the requirements for the
external signal sources, including recorded calibration of a system to be used for speech
speech material playback and microphone for audiometry, we must look at the individual
live voice. When using a microphone, the components of the system and how they might
manufacturer of the audiometer is required to influence the reliability of the test person’s
specify how it should be used. response. For convenience, we can consider
Similar to the frequency response toleran- the system used for the purposes of performing
ces, the requirements for harmonic distortion speech audiometry to consist of three main
produced at the earphone are given in IEC components: first, the speech material itself,
60645-1 (2012) and for the loudspeaker in IEC which is, in general, in a recorded format;
60645-2 (1993). For both earphones and loud- second, the equipment used to control the
speakers, harmonic distortion is measured using presentation characteristics of the speech mate-
a pure-tone signal that is 9 dB above the rial, for example the speech level; third, the
indicated reference level on the signal indicator. method of delivery to the test subject: earphone
All types of speech audiometers are re- or sound field.
quired to provide speech-weighted random It must be ensured that not only are we
noise for speech masking. The spectrum level delivering test material to the test person at the
of the weighted noise delivered by the loud- appropriate level but that the speech material
speaker at the reference point in the sound-field being used is of an appropriate quality and that
test environment shall be constant from 125 Hz the integrity of the recorded material is main-
to 1,000 Hz, should drop by 12 dB/octave from tained when delivered to the test person’s ear or
1,000 to 6,000 Hz level, and should not exceed ears. The general requirements for recording
the 6,000 Hz level at frequencies above 6,000 the speech material are specified in ISO 8253-3
Hz. The level shall vary by no more than
5 (2012). To facilitate the qualification (i.e., the
dB between the frequencies 250 and 4,000 Hz signal meets specified requirements) and cali-
and
8 dB between 4,000 and 6,300 Hz. bration of a speech audiometry system, each
These requirements also apply to the uncor- copy of the recording is required to contain
rected SPL produced by an earphone when three components. Not only is it required to
measured in the appropriate acoustic coupler provide a good-quality recording of the speech
or ear simulator. material (the first component), but also (sec-
ISO 8253-3 (2012) specifies basic methods ond) a signal to be used for the calibration of the
for speech recognition testing to ensure com- equipment and (third) additional signals for
parability between different test procedures, testing the frequency response of the equipment
including the use of different languages. It to be used for delivering the material to the
also specifies the requirements for the compo- subject. These latter two signals may be used for
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 353

testing the equipment used for playback and the equivalent SPLs. If speech material is used of
recording. alternative durations (i.e., less than 15 seconds)
To maintain the integrity of a recording, all or a frequency weighting other than C-weight-
copies of specific speech material, together with ing, then correction factors should be applied to
its accompanying calibration signals and equip- compensate for this.
ment test signals, shall be made from the ISO 8253-3 (2012) requires that each copy
original reference recording. Any changes of a speech-material recording be supplied with
made to a reference recording will require documentation that specifies the level of the
new validation. calibration signal and the characteristics of the
Speech tests in noise representative of calibration signal and other signals provided for

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
everyday communication provide valuable in- the verification of the test equipment frequency
formation that cannot be obtained from speech response. These characteristics should include
in quiet. If the speech audiometry is intended to the duration, frequency spectra, and levels
be performed in the presence of competing relative to the speech level of the calibration
noises, the noises to be used should be included signal. The documentation should specify the
in the same reference recording. This can be on characteristics of any competing noise signals
the same or separate track depending on the and related calibration signals that may be
requirements of the test method to be used. provided with the speech-material recording.
These recordings shall also make provisions for In addition to the speech-material recording
calibration. requirements, ISO 8253-3 (2012) also specifies
The speech material should be appropriate the requirements for the recording equipment.
for the test method to be used and the ability of Before routine use, the speech-material record-
the test subject to participate. For example, ing should be validated using audiometric
when testing a child, the vocabulary and lan- equipment complying with IEC 60645-2
guage used in the test material should be (1993). This procedure also forms part of
appropriate to their age and ability. Test meth- ISO 8253-3 (2012).
ods that require repeat testing should be suffi-
cient to minimize the risk of familiarization or
being learned by the test person from repetitive THE CALIBRATION AND
presentation. The calibration signal should VERIFICATION OF A SPEECH
consist of a band of weighted random noise AUDIOMETRY SYSTEM
centered at 1 kHz with a bandwidth of one- To ensure the reliability of the results obtained
third octave or a 1-kHz frequency-modulated when carrying out speech audiometry, ISO
tone with a bandwidth of at least one-third 8253-3 (2012) recommends a scheme that
octave. This should be modulated using a consists of three stages of performance checks
sinusoidal or triangular signal with a repetition and calibration.
rate within the range 4 to 20 Hz (see IEC
60645-2 1993). This signal should have a
duration of not less than 60 seconds. The Stage A: Routine Checking and
calibration signal shall be within
0.5 dB of Listening Tests
the average speech level of the speech material, Stage A recommends routine checks and lis-
and its level should be measured using the same tening tests that are designed to ensure that the
weighting as for the speech level. The signals equipment is functioning properly and that
used for verification of frequency response there have been no noticeable changes in its
should consist of one-third-octave bands of calibration. It also recommends that these tests
white noise centered at the preferred one- are performed in conditions of normal use. The
third-octave frequencies in the range 125 to guidance is that stage A checks should, where
8,000 Hz as specified in ISO 266 (1997). Each practicable, be performed before the equipment
signal should have a duration of not less than 15 is used each day. The recommendation is that
seconds. It is recommended that speech mate- the equipment, including leads, connectors, and
rial recordings are made using C-weighted earphone cushions, should be examined for
354 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

wear or damage and if necessary replaced. parts of IEC 60645-2 (1993) and consist of
Serial-numbered parts should be checked to measuring the frequency response of the equip-
make sure they are still the same to avoid errors ment and its output transducers, the output
arising from events such as earphones being levels of the transducers, the attenuator steps
changed. Where appropriate, the equipment (IEC 60645-1 2012), the harmonic distortion,
should be cleaned. the masking noise levels, and the headband
With the equipment switched on and force of the transducers. These electroacoustic
warmed up for at least 5 minutes, any routine tests are very similar to those specified in ANSI
setup procedure as required by the manufactur- S3.6 (2010).
er should then be performed. For sound-field

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
testing, it should be ensured that any loud-
speaker is in the correct position and that the Stage C: Basic Calibration Tests
reference position is clearly marked at the Basic calibration tests are performed to ensure
correct position. Using an individual who is that all the audiometric equipment used to
known to have normal hearing thresholds, it perform the specific test and the test environ-
should be confirmed that the output of the ment complies with all relevant requirements of
audiometer is approximately correct by listening IEC 60645-2 (1993) and ISO 8253-3 (2012).
to the speech test signal at a low level and that Stage C tests are not required routinely if stage
the ambient noise levels are acceptable. Ensure A and stage B tests are correctly performed and
that there is no hum or unwanted noise coming are necessary when a serious fault has occurred
from the equipment. and been rectified, which may have adversely
The speech test material should be listened affected the systems performance, or after a
to at a higher level to ensure that there is no long period when the required performance of
noticeable distortion or interfering noise. For the equipment cannot be assured. The standard
this check, a hearing level for speech of 60 to does recommend that equipment should under-
70 dB is recommended. Correct functionality go stage C tests at least every 5 years.
of the attenuators over their range should also
be checked. These checks should be performed
with all the output transducers to be used for the INSTRUMENTATION REQUIRED
subsequent audiometry and, when possible, FOR CALIBRATION
performed by the same person each time. If For the purposes of this article, we will consider
the test method requires a subject’s talk-back the instrumentation required to carry out stage
system, it should be checked that this is func- B (periodic electroacoustic tests) on the speech
tioning correctly. audiometer and its associated components.
Measurements made on the acoustic outputs
of both earphones and loudspeakers will be
Stage B: Periodic Electroacoustic Tests included.
Stage B tests require acoustic measurements to All basic electroacoustic measurements to
be made, and the standard recommends that be performed for the purposes of calibration and
these measurements are made at intervals be- verification require either a sound level meter
tween 3 and 6 months. ISO 8253-3 (2012) (SLM) or measuring amplifier that complies
does, however, allow for different stage B test with the relevant parts of IEC 61672-1 (2002).
intervals provided that stage A checks are Also included are any microphones, depending
performed regularly, but recommends a maxi- on which type of output transducer or trans-
mum interval of 12 months. Stage B tests ducers are used; a one-third octave filter set,
should include a stage A check that, if per- which complies with relevant parts of IEC
formed prior to the electroacoustic measure- 61260-1 (1995); a distortion meter, although
ments, should identify any issues that need to be filters can be used for measuring harmonic
addressed before continuing. distortion; and the appropriate acoustic cou-
The required electroacoustic tests are per- plers or ear simulators required for measure-
formed to ensure compliance with the relevant ments of earphone outputs. Each coupler or
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 355

simulator will need to meet the specifications of Masking sound characteristics

the relevant part of IEC 60318-1 (2009), IEC Output SPLs and frequency responses for all
60318-3 (1998), IEC 60318-4 (2010), IEC transducers and methods of delivery to be
60318-5 (2006), ANSI S3.25 (2009), and used (including live-voice microphone if
ANSI S3.55 (2014). If distortion measure- appropriate)
ments are to be made, it is more convenient if Test subject response talk-back system
a distortion meter is used, although a filter set
can be used for this purpose. Some SLMs and For many speech test methods, the re-
measuring amplifiers incorporate additional sponse from the test person to a particular
functionality that can remove the need for test item is spoken. When this test method is

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
some of the instrumentation detailed previous- used and the tester is not in the same room as
ly. For example, some SLMs and measuring the test subject, a talk-back system is required to
amplifiers have an integral filter set or spectrum relay the spoken response to the tester.
analyzer that can meet the requirements of
calibration and verification, and some will in-
clude harmonic distortion measurement. For CALIBRATION PROCEDURE
sound-field calibration, it is useful to have an Unlike the calibration of a pure tone audiome-
extension microphone cable, so that measure- ter, which provides facilities to make adjust-
ments can be made without significantly dis- ments to specific pure tone frequencies or
turbing the sound field, and a tripod to support narrowband signal levels, a speech audiometer
the microphone at the reference point. A sound will, in general, not provide facilities for adjust-
calibrator is required with appropriate adaptors ing the frequency response of broadband signals
for the microphones to be used. This will help such as speech and speech-weighted noise. It is
ensure the accuracy of the measurements made. the responsibility of the manufacturer of the
If suitable recorded test signals are not available, speech audiometer to ensure that the frequen-
either on the sound material recording or from cy-response requirements of IEC 60645-2
an alternative recording, a signal generator (1993) and ISO 8253-3 (2012) are met, or
capable of producing the test signals will be may be met, when using specified transducers
required. for delivering the material to the test subject
and when using external components such as
devices for playing recorded speech material,
PARAMETERS AND live-voice, and patient-response microphones
FUNCTIONALITY REQUIRING and speakers for sound-field audiometry. The
CALIBRATION AND/OR FUNCTION speech audiometer should provide facilities for
VERIFICATION the adjustment of overall signal levels and for
The parameters that may require calibration or the appropriate adjustment of calibration sig-
verification of function and performance de- nals provided with recorded speech material.
pend on both the functions available on a Prior to making any electroacoustic meas-
specific speech audiometer and the actual func- urements, a stage A check should be performed
tions and features to be used for specific test to confirm the correct functionality of the
materials and test procedures. These may speech audiometer and any associated equip-
include: ment to be used with it to perform the audiom-
etry. If problems are identified during the
Overall frequency response of the speech check, they should be rectified before proceed-
audiometer ing. Any setup of the equipment required and
Loudspeaker output level and frequency specified by the manufactured should then be
response performed and the equipment allowed to warm
Output level scale up to its normal operating temperature before
Attenuator characteristics carrying out measurements. It must be ensured
Scale reference and output level that the equipment being used to make the
Harmonic distortion measurements is appropriately calibrated prior
356 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

to use, and a reference sound calibrator should field environment, shall be performed using a
be applied to the measurement microphone to free-field microphone located at the reference
confirm calibration. Any corrections associated point, as shown in Fig. 1. For the purposes of
with the measurement equipment should be sound-field audiometry, the reference point is
recorded and applied as appropriate to measure- the midpoint of a straight line connecting the
ments made. ear canal openings of the listener when posi-
Measurements should then be made to en- tioned in the listening position in the sound
sure that all relevant requirements of the standards field. For sound-field speech audiometry, the
are met. Test requirements to demonstrate con- calibration should be performed within the
formity to relevant parts of IEC 60645-1 (2012) same environment to be used to perform the

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
are described within that standard and the general audiometry.
requirements for the specification, testing, and
calibration of speech audiometers, together with
frequency response requirements, are described in VALIDATION OF SPEECH
IEC 60645-2 (1993). MATERIAL RECORDINGS
For the purposes of calibration, all the ISO 8253-3 (2012) requires that speech-mate-
measurements of the acoustic output of an rial recordings be validated to obtain reference
earphone shall be performed using the acoustic test data for the material being used. ISO 8253-
coupler or ear simulator, complete with its 3 (2012) specifies methods by which these data
pressure microphone, which is appropriate for can be obtained. It also details how repeatability
that specific earphone. Measurements of the of results can be determined and test–retest
acoustic output of a loudspeaker, in a sound- reliability.

Figure 1 Sound-field measurement arrangement.

 AUDIOMETRIC CALIBRATION/FROST, LEVITT 357

SPEECH INTELLIGIBILITY INDEX therefore do not have access to routine speech

American National Standards provide a meth- testing methods. In these situations, it may be
od for computing a physical measure that is possible to use logatoms or nonsense words.
highly correlated with the intelligibility of The effective use of speech test material
speech, as evaluated by speech perception tests assumes that the test subject has some under-
given a group of talkers and listeners (ANSI standing of the language, particularly at the
S3.5 1997).This measure is called the Speech phoneme level, and the grammar used, even
Intelligibility Index and is calculated from when nonsense words are used. Many speech
acoustical measurements of speech and noise. test methods require that the test subject repeats
what they have heard and his or her response is

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
noted or scored by the person conducting the
LIMITATIONS OF CURRENT test. This is achieved either by the tester
STANDARDIZATION AND listening to the test person in close proximity
CALIBRATION PROCEDURES or by having the spoken response relayed via a
International standards published by IEC and talk-back system. The quality of the talk-back
ISO often have similar or equivalent standards relay system and the ability of the tester to
published by ANSI. This, however, is not the accurately hear the response is fundamental to
case for standards directly relating to speech the reliability of the test results, but references
audiometry. Those who normally refer to ANSI to the requirements of this type of system in
publications can in their absence use IEC and current standards are extremely vague. In prac-
ISO standards as an alternative. It would be tice, the microphone used to detect the test
helpful for the purposes of international stan- subject’s voice is often some distance away from
dardization, however, if all appropriate ISO and them and located in a far-from-ideal position.
IEC standards were published as parallel or, The resulting microphone input signal may
ideally, harmonized ANSI standards. often be of poor quality with a significantly
Because of the difficulties associated with degraded frequency response. The signal from
the process of recording speech material, so that the microphone is then relayed through an
it both maintains the integrity of the original amplifier to either earphones or a loudspeaker,
speech and meets the requirements of ISO which may be of relatively poor quality. In
8253-3 (2012), audiologists are sometimes in practice, the test person’s response as heard by
a position where they are unable to access the tester can be significantly different from
appropriate material. In such situations, they their actual response. Test results can therefore
may revert to using recorded material that, in be adversely affected, thus making testing ex-
some cases, may have been recorded a consid- tremely unreliable. The talk-back relay system
erable time ago. These older recordings are can be an integral part of the test system, and it
often analog and rather than being taken di- can be strongly argued that there is a need for its
rectly from a reference recording, as required by performance to be specified and standardized.
ISO 8253-3 (2012), they are multiple-genera- When test methods are used where the test
tion copies and are of poor quality, with a subject is required to repeat what has been
significantly reduced frequency response, intro- heard, test results and scores are reliant on
duced noise, and unreliable calibration signal. the subject’s ability to articulate and repeat
These recordings sometimes include vocabulary accurately what has been heard. This method
or grammar that is no longer in current use. of testing relies on normal speech production by
Results obtained using this material are there- the test subject. Those with severe hearing loss
fore extremely unreliable and this material may have impaired speech. These methods also
should not be used. Even those recordings rely on the tester’s ability to discriminate what
that have been rerecorded digitally and have the test subject has said; if the subject’s speech is
been digitally cleaned often contain significant- impaired or if the subject has a strong accent or
ly degraded material and should not be used for dialect, the response may be incorrectly heard
reliable testing. Also some countries do not by the tester, resulting in a reduced score on the
have speech material in their local language and speech test. These speech test methods can
358 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

therefore be as much an evaluation of the test time-varying characteristics such as pauses,

subject’s ability to speak and the tester’s ability phrasing, and breaks between words and sen-
to discriminate speech as they are a measure of tences. The problems associated with the mea-
the test subject’s ability to discriminate speech. surement of speech are further exacerbated by
This potentially makes this type of speech test the characteristics of the measurement instru-
method extremely unreliable, and standardiza- mentation, where appropriate measurement
tion and calibration can do little to address this. time constants and frequency weightings
With forced-choice testing, the choice have to be considered. For most applications,
options, whether written or graphic, must be the use of current measurement techniques
very carefully selected. A test person may select such as peak and average SPL are considered

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
an option because it has a similar sound to what appropriate and have uncertainties in measure-
they may have heard, and this may not neces- ment that are consistent with other relevant
sarily reflect their ability to discriminate a parameters.
particular word. The speech material and the
choice options must be very carefully selected
to achieve the required outcome of a specific FUTURE DEVELOPMENTS: THE
test. This process is extremely difficult to IMPACT OF DIGITAL TECHNOLOGY
standardize, and those developing tests of Major technological advances affect both meth-
this type must consider their reliability. Some ods and mind-set. The invention of the elec-
currently available material may give mislead- tronic amplifier ushered in an era in which
ing results. electronic instruments became the tools of the
Speech materials may be used to see how trade. A different mind-set was needed for early
an individual may perform and respond to hearing professionals to make the change from
different input levels of speech for rehabilita- tuning forks and ear trumpets to audiometers
tion purposes. These tests may be performed in and hearing aids. Electronic technology at that
a sound-field environment using standardized time provided a practical means for measuring
speech material presented at different levels the frequency response of amplifiers, filters, and
using the audiometer attenuator. With soft other audiometric equipment. Measurements
spoken speech and raised voice speech, howev- of the impulse response provide the same
er, the spectral and temporal characteristics of information much more rapidly but in a differ-
the speech normally change as the level of the ent mathematical form, which is difficult to
spoken voice changes. These changes in char- interpret for a mind-set familiar with tuning
acteristics are not replicated when the level of forks. Impulse response measurements were
standardized recorded speech material is var- difficult to implement with primitive electron-
ied, and this method of evaluation does not give ics and did not receive much attention during
a true indication of how a hearing aid user may the early electronic era.
perform in these different listening situations. The limitations of early electronic instru-
When recorded speech material is to be used mentation and concomitant constraints on the
for this type of testing, it should be specifically mind-set at that time resulted in a frequency-
recorded and standardized for this purpose and based measure of auditory sensitivity, the au-
should reflect the changes that result from diogram, which became a basic, standard mea-
speaking at different levels. It can be argued sure of hearing. The audiogram, however, is an
that there is a need for more specific measures incomplete measure in that it provides no
of speech level but due to the complex nature of information on the temporal characteristics of
its characteristics, measurement parameters hearing. Recent research has shown that re-
and methods for their measurement are ex- duced temporal resolution is an important
tremely difficult to define or specify. Speech sequela of the aging process, and it is a major
alone includes variations in its spectral, inten- factor contributing to poor speech understand-
sity, and temporal characteristics, but the mea- ing in noise for both normally hearing and
surement of continuous speech presents hearing-impaired people. Practical, clinical
additional challenges associated with other methods for the measurement of temporal
 AUDIOMETRIC CALIBRATION/FROST, LEVITT 359

resolution need to be developed and part of the message, yet other pauses may
standardized. indicate that the talker has simply stopped
The field of audiology, in common with speaking.
many other fields, is currently experiencing the Digital techniques allow for true rms levels
impact of digital technology. It is very impor- to be obtained with little difficulty (as opposed
tant in developing new methods of measure- to the difficulty of obtaining true rms with
ment and associated standards to think in terms analog instrumentation). A problem with rms
of what digital technology can do that the older measurements of speech level is that the mea-
analog technology could not do. There is evi- sured level depends on the averaging time, and
dence that this change in thinking is taking there are conditions in which two very different

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
place, but in a two-stage process. The first averaging times could be appropriate depending
digital hearing aids, for example, implemented on the application. Consider a word with a weak
the same basic methods of amplification as sibilant in word-final position that trails off very
analog hearing aids, but with greater precision slowly. If the averaging time includes the entire
and improved quality. The result was an im- sibilant as produced by the talker, including the
proved hearing aid that was only slightly better last portion that is not audible to the listener,
than a conventional analog hearing aid. The then one value for the level of the word will be
next generation of digital hearing aids, howev- obtained. If the averaging time includes only
er, introduced novel methods of signal process- that portion of the sibilant that is clearly audible
ing that could not be implemented in an analog to the listener, then another value for the level
hearing aid, such as adaptive methods of feed- of the word will be obtained. A similar problem
back cancellation. These new methods have occurs in measuring the level of conversational
resulted in digital hearing aids that are substan- speech in the presence of background noise.
tially superior to analog hearing aids. Should the averaging time include that portion
Digital technology has had a similar impact of the signal where speech is present but
with respect to instrumentation for the mea- completely masked by noise? An approach to
surement of speech levels. Digital instruments this problem is to obtain the cumulative distri-
soon replaced the VU meter as the standard bution of instantaneous values of the speech
instrument for measuring speech levels in tele- plus noise and define the speech level in terms of
phone and broadcast networks. Initially, a digi- a percentile that is well above the noise level,
tal simulation of the VU meter was used before with a correction based on the difference in
digital instruments were introduced that mea- shape between observed cumulative distribu-
sured speech in a completely different way. Peak tion and one obtained previously for speech in
program meters (PPMs) are now widely used in quiet. Standards need to be developed for
the broadcast industry. There are now several audiologic applications in which these consid-
variations of this meter, such as True PPM, erations are taken into account.
Quasi PPM, Sample PPM, among others, and The development of new digital products
each version can be implemented by means of or new digital methods is far more rapid than in
software within the same instrument. Digital the past with analog technology. One reason is
rms meters have largely replaced VU meters in that software development is both more rapid
research studies, although there are applications and more efficient than hardware development.
in which VU meters are still used.18 The demand for novel digital products is also far
Digital techniques allow for many different greater than in the past. This has important
ways of measuring speech level to be imple- implications for standardization. The develop-
mented, and the problem now is defining ment of standards is a relatively slow process. It
exactly what needs to be measured. Speech is was a major problem in the predigital age to
a complex signal in which some silences are an develop standards fast enough to keep up with
integral part of the speech signal, such as the technological advances. It is likely to be an even
silence before voice onset in a stop consonant, more difficult problem in the digital future with
whereas other silences indicate a pause. Some its increasingly rapid development of new
pauses may convey meaning and are therefore a methods and products.
360 SEMINARS IN HEARING/VOLUME 35, NUMBER 4 2014

The recent explosive growth in video 5. Fletcher H, Allen JB. ASA Edition of Speech and
speech communication by means of the Internet Hearing in Communication. Melville, NY: Acous-
or mobile phones is a case in point. New digital tical Society of America; 1995
6. Fletcher H, Steinberg JC. Articulation testing
devices for video speech communication are
methods. Bell System Technical Journal 1929;
being introduced at a rapidly increasing rate. 8(4):806–854. Available at: http://www3.alcatel-
The quality of the video images on these digital lucent.com/bstj/. Accessed on September 22,
devices is often quite poor. People with hearing 2014
loss as well as seniors with normal hearing are 7. Hudgins CV, Hawkins JE, Karlin JE, Stevens SS.
less able to understand video speech signals with The development of recorded auditory tests for
poor image resolution, poor contrast, and/or measuring hearing loss for speech. Laryngoscope

This document was downloaded for personal use only. Unauthorized distribution is strictly prohibited.
1947;57(1):57–89
slow frame rates. Appropriate standards need to
8. Hirsh IJ, Davis H, Silverman SR, Reynolds EG,
be developed so that seniors and people with Eldert E, Benson RW. Development of materials
hearing loss can use these new methods of for speech audiometry. J Speech Hear Disord 1952;
communication. This is an urgent problem 17(3):321–337
because many people with hearing loss as well 9. Carhart R. Tests for selection of hearing aids.
as seniors with normal hearing are not benefit- Laryngoscope 1946;56(12):780–794
ting from these devices despite the potential to 10. Levitt H. The phonograph, telephone, hearing
aids, and the development of speech audiometry.
communicate more efficiently and more conve-
In Levitt H, Pickett JM, Houde RA, eds. Sensory
niently with well-designed video speech Aids for the Hearing Impaired, Part II, Appendix.
systems. New York, NY: IEEE Press; 1980
Digital technology can also be disadvanta- 11. Romanow FF. Methods of measuring the perfor-
geous to people with hearing loss in unexpected mance of hearing aids. J Acoust Soc Am 1942;
ways. Digital mobile phones generate signifi- 13(1):294–304
cant interference in hearing aids that are close to 12. Zwislocki JJ. An Acoustic Coupler for Earphone
Calibration. (Special Report LSC-S-7). Syracuse,
the digital device, as when a mobile phone is
NY: Syracuse University, Laboratory of Sensory
held close to an ear with a hearing aid. Stand- Communication; 1970
ards for specifying acceptable levels of interfer- 13. Zwislocki JJ. An Ear-like Coupler for Earphone
ence in hearing aids have been developed (IEC Calibration. (Special Report LSC-S-9). Syracuse,
60118-13 2011; ANSI C63.19 2011).19,20 NY: Syracuse University, Laboratory of Sensory
Improved methods of wireless communi- Communication; 1971
cation are constantly being introduced that 14. Dillon H. Hearing Aids. 2nd ed. Turramurra,
Australia: Boomerang Press; New York, NY:
affect wireless interference in hearing aids,
Thieme Publishing Company; 2012
including new features in modern mobile 15. Zwislocki JJ. Auditory Sound Transmission: An
phones, and efficient ongoing updating of these Autobiographical Perspective. Mahwah, NJ: Law-
standards is a challenge for the standardization rence Erlbaum Associates; 2002
process. 16. Chinn HA, Gannett DK, Morris RM. A new
standard volume indicator and reference level.
Proc Inst Radio Eng 1940;28:1–17
REFERENCES 17. Killion MC. Comparison of vu-meter-based and
rms-based calibration of speech levels. J Acoust Soc
1. Martin M. Speech Audiometry. 2nd ed. Oxford, Am 2009;126(3):EL97–EL99
UK: Wiley-Blackwell; 1997 18. Lobdell BE, Allen JB. A model of the VU (volume-
2. McArdle R, Hnath-Chisolm T. Speech Audiome- unit) meter, with speech applications. J Acoust Soc
try. In: Katz J, Medwetsky L, Burkard R, Hood L, Am 2007;121(1):279–285
eds. Handbook of Clinical Audiology. 6th ed. 19. Levitt H. The nature of electromagnetic inter-
Baltimore, MD: Lippincott, Williams & Wilkin; ference. J Am Acad Audiol 2001;12(6):
2009:64–79 322–326
3. Campbell GA. Telephonic intelligibility. Philos 20. Levitt H, Kozma-Spytek L, Harkins J. In-the-ear
Mag 1910;19:152–159 measurements of interference in hearing aids from
4. Fletcher H. Speech and Hearing. New York, NY: digital wireless telephones. Semin Hear 2005;
Van Nostrand; 1929 26(2):87–98
Copyright of Seminars in Hearing is the property of Thieme Medical Publishing Inc. and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.