tvi‘., veterans administrationanna k. nabelek, ph . d. amy m. donahue, m.a. tomasz r. letowski,...

tVi‘., VeteransAdministration

Journal of Rehabilitation Researchand Development, Vol . 23, No . 1, 1986pages 41-52

Comparison of Amplification Systemsin a Classrooms

Abstract — "Listening systems" are used for hearing impairedlisteners as an alternative to public address systems (PA) used forthe general public . These listening systems allow individual con-trol of sound pressure level and minimize the effects of backgroundnoise and room reverberation . Three listening systems, based on(i) an audio induction loop (AL), (ii) frequency modulation of radiofrequencies (FM),and (iii) modulation of infrared light (IR) werecompared among themselves and with a PA system in a medium-size classroom. Listening groups were (i) normal hearing, (ii) hearingimpaired, (iii) hearing aid users, and (iv) elderly . Word identificationstores were obtained with the Modified Rhyme Test at two condi-tions: with a babble of 12 voices at a speech-to-noise ratio (SIN) of+8 dB, and without the babble at SIN of +20 dB . Analysis ofvariance indicated that the main effects of systems, groups androom S/N were significant . Also significant were interactions ofsystems by groups, and systems by SIN . For all groups, the threelistening systems provided better scores than the PA system . Itcan be concluded that all three of the tested listening systems aresuitable for listeners with various degrees of hearing losses.

INTRODUCTION

Perception of speech in rooms is affected by several factorsthat were described by Knudsen in 1929 in the following way:" . . .the resulting percent articulation in any specified auditoriumcan be estimated by the following equation:

Percentage Articulation = 96 k Q k r k n ks

where k Q is the reduction factor owing to the inadequate loudnessof the speech, k r the reduction or distortion factor owing to re-verberation, k n the reduction factor owing to noise, and k s thereduction factor owing to the shape of the room" (1).

Communication involves the transmisstion of acoustic wavesfrom a source, through an environment, to a receiver . Knudsen'smodel refers primarily to the effects of environment, with the ex-ception of kQ which refers to both the source output level and tothe distance between source and receiver . To describe Knudsen'spercentage articulation more fully, two factors should be added tothe model : k t describing the talker or other source of speech andkp describing the perceptual abilities of the listener .

41

ANNA K . NABELEK, Ph . D.AMY M. DONAHUE, M .A.TOMASZ R . LETOWSKI, Ph . D.

Dept . of Audiology & Speech PathologyUniversity of Tennessee, Knoxville457 South Stadium AnnexKnoxville, TN 37996-0740

'This research was supported by a grantfrom the National Institute of Neurolog-ical and Communicative Disorders andStroke, U . S . Public Health Service No.R01 NS 12035.

NOTE : Requests for reprints should besent to : Anna K . Nabelek, Ph . D ., Depart-ment of Audiology & Speech Pathology,University of Tennessee, Knoxville, TN37996-0740 .

42

NABELEK et al . : Amplification Systems in a Classroom

It has been well documented, i .e ., by Nabelek andPickett in 1974 (2), and by Finitzo-Hieber and Tillmanin 1978 (3), that for good speech perception, hearingimpaired listeners require better speech-to-noiseratios (S/N) and shorter room reverberation thannormal hearing young adults . In addition, hearingimpaired listeners need sound pressure levels(SPL)higher than normal hearing listeners do, to com-pensate for their hearing loss.

Thus there are three conditions necessary to as-sure a good listening environment for hearing im-paired listeners : (i) adequate SPL, (ii) low backgroundnoise, and (iii) short reverberation . It is expectedthat modern technology will be able to fulfill thesethree requirements . Amplification of the sound levelcan be accomplished at two points ; at the source,or at the listener . Sound from the source can beamplified and transmitted to the room at levelshigher than could be obtained using only the livevoice. Electroacoustic technology has developedsufficiently so that the reproduced sound need notlose the "live voice" quality . Most modern facilities,such as auditoriums and places of worship, useamplification systems, sometimes called publicaddress systems (PA) . A PA system consists of amicrophone and one or more loudspeakers . Propermicrophone placement is critical for high fidelityreproduction. If more than one microphone is needed,the expertise of an audio engineer is often required.The loudspeakers deliver amplified sounds to a lis-tening area. Frequently, loudspeakers are locatedon each side of a stage, podium, or altar, or theloudspeakers may be located above the stage . Awell-designed PA system should deliver an evenSPL over the entire listening area . The SPL is set byan audio technician to produce comfortable listen-ing for a normal-hearing audience. This level may betoo low for hearing impaired listeners . Moreover,since the loudspeakers are located at a distancefrom the listeners, the arriving sounds are mixedwith background noise and degraded by room re-verberation in varying amounts depending on a lis-tener's location in the room . In effect, the listeningconditions may not be fully satisfactory for a hearingimpaired listener.

Hearing impaired listeners can use personal am-plifying devices, such as hearing aids . The hearingaid amplifies sound to the level required by theindividual listener and may also shape frequencyresponse to compensate for different degrees ofhearing loss at various frequencies . However, thehearing aids indiscriminately amplify both the wantedsounds and background noise, being unable to dif-ferentiate between the direct, non-reverberant and

reverberant signals . Speech enhancement in noisyand reverberant listening conditions has not yetbeen accomplished by presently available technol-ogy for a wearable hearing aid.

At this time, the only known practical remediesare a proper room design which maintains a lowbackground noise level and short reverberation time,or use of a listening system . A listening system de-livers speech signals from the source directly to areceiver placed at the listener's ear, thereby evadingthe problem of background noise and room ' rever-beration . The gain control in the receiver allowsindividual setting of the SPL.

The following five systems have been thus farused for delivering speech signals to hard-of-hearinglisteners:

1. Hardwired;2. Audio induction loop (AL);3. Amplitude modulation (AM) of radio signals;4. Frequency modulation (FM) of radio signals ; and5. Modulation of infrared (IR) waves.Several publications have indicated that listening

systems can be beneficial for listening even if lessthan optimal conditions exist . Hawkins, Fluck, andVan Meter in 1982 described an FM system installedin a large auditorium for the general public and forhearing-impaired listeners (4) . Speech perceptionscores were not obtained but the authors reportedthat 90 percent of the patrons who responded to aquestionnaire indicated satisfaction with the equip-ment . Vaughn in 1983 and Williams in 1984 describedfive listening systems used in auditoriums forhearing impaired listeners, discussing applications,advantages, disadvantages, and approximate cost(5, 6). They did not compare speech perceptionamong the systems . Bankoski and Ross in 1984compared speech discrimination in an auditoriumfor normal-hearing and hearing impaired listenerswith a PA and an AM/FM system (7) . When subjectswere sitting in the center and rear sections of theauditorium, the FM system yielded speech percep-tion scores which were significantly better for bothgroups of subjects . While the authors indicatedthat similar results probably would have been ob-tained with any other listening system, the testingwas limited to the FM system.

The goal of this study was to compare speechperception with PA, AL, FM, and IR systems . (Thehardwired and AM systems were not tested becausethey are becoming obsolete ; they are, however,described here to provide thorough comparison ofall the systems which have been in use and may beencountered in public places .) The systems were

r

Y

43

Journal of Rehabilitation Research and Development Vol . 23 No . 1 1986

°

-s

tested without microphones ; the test speech wasdelivered from a tape recorder . All testing wasperformed in a medium-size classroom using lis-teners with various degrees of hearing loss. Onegroup of listeners wore hearing aids.

Listening Systems

Hardwired — The hardwired listening systemsemploy a direct wire connection between a micro-phone positioned at the signal source location andan earphone or hearing aid located at the listener'sear. These installations usually provide a very goodquality signal, and are not susceptable to electro-magnetic interference . They are easily designed foronla!l rooms but can become complicated and ex-pensive in large auditoriums . They are prohibitivelycumbersome and of limited value for public placeswhere full mobility must be obtained . Cable break-age is the main maintenance problem.

Audio induction loop -- Induction loop systemshave been used as permanent installations orportable facilities for both group and individuallisteners . The essential system component is amultiturn loop of wire encircling a room and con-nected to the output of an audio power amplifier.The signal fed into the amplifier can come from amicrophone, tape recorder, or similar source . Theelectric current passing through the loop creates aroom magnetic field that varies in strength andfrequency according to the speech signal . Thesignal carried by the alternating magnetic field hasto be converted back into an electric signal whichis used to produce an acoustic signal that is deliv-ered to the listener's ear. The magnetic-to-electricconversion can be achieved in a small pickup loop(also called ate!e!oop)connected to the input of areceiver, or in a pickup coil of a hearing aid switchedto the "T" (telephone) position.

The layout and dimensions of the transmittingloop depend on the shape and size of the room . Inorder to obtain a strong and uniform magnetic fieldin a whole room, the loop must be individuallydesigned. Loops for use in small areas are simpleand commercially available ; the loop for a largeh8\l can be complicated, especially when someconstructional restraints exist.

The main interferences are caused by audio andelectrical equipment used in the same or adjacentrooms . The most frequent sources of interferencesare devices such as light dimmers, metal detectors,fluorescent lights, and other current-carrying,magnetic-field-generating magnetic loops including

AC power cables and the coils of electric motors.High levels of existing or projected interferences ina building can preclude the use of AL systems.

AM and FM systems—In both AM and FM systems,the audio signal is transmitted by radiofrequencycarrier . In AM transmission, the amplitude of thecarrier is changed proportionally to the changingamplitude of an audio signal . In FM transmissionthe amplitude of the carrier remains constant, andits radio frequency is made to change proportionallyto the changing amplitude of an audio signal.

Aradinfrequancy listening system, regardless ofits mode of modulation, consists of a microphoneand transmitter unit with an antenna, and a receiverconnected to an earphone ore hearing aid . Themicrophone and transmitter unit imposes thespeech or other audio nignel onto the electro-magnetic waves of radio frequency which areradiated by an antenna into the environment . Thereceiver converts radiofrequency waves back intoan audiofrequency electrical signal which is usedto produce an acoustic signal that is delivered tothe listener's ear . The maximum distance of trans-mission depends on the radio transmitter power,the type of transmission, and the receiver's sensi-tivity.

For an AM listening system, as a result of FCCregulations limiting the radiated field strength, anantenna has to be specially designed . in many AMsystems, the antenna is a one-turn loop of wireencircling the room . Its installation may be asdifficult as the installation of a multiturn loop foran audio induction system . The typical radio carrierfrequency is in the ordinary AM radio band of 525-1610 kHz or below . The receiver can be a commer-cially available AM pocket radio or a special fixed-frequency receiver. The range of transmission,limited to a maximum of 15 meters, is usuallyadequate for covering areas of small theaters andchurches and does not create problems with unin-tended "broadcasting" outside the building . How-ever, the nature of AM transmission can createproblems when the goal is to achieve a homo-geneous field in a\arge hall . Additionally, AM sys-tems are very susceptible to high frequency electricalinterferences generated within the same building(e .g. by electric motors) and by strong externalsources.

Generally, AM listening systems do not performwell in buildings with aubetantial amounts of struc-tural steel.

FM systems usually operate at higher radiofrequencies than do AM systems . FM's shorter

44

w~BEUEKeueLAmplification Systems inaClassroom

radio waves are less susceptible to the absorptionphenomenon caused by structural ateel in the build-ing. Furthermore, these frequencies are easilyradiated into a room by short antennas . Thetrans-mitting strength of an FM system allows for goodreception within very large rooms. However, thesystem can radiate outside the enclosure, whichoanprndune bothinterf*nenooaand^'eaveedropping"problems . The main advantage of FM systems overAM systems is that FM transmission is less affectedby interfering aourooe, such as radio signals, sparks,lightning, electric motors, etc.

Until 1982, FCC regulations restricted the use ofFM listening systems to educational institutions.As a result, the AM systems were promoted in the1970'eaaan inexpensive listening aid forhard-of'hearinOpeop!minpub!iop!aoea ' butsinoe18M2thoyhave been losing their appeal in comparison with theFM systems. The FM systems, comparable in priceto AM systems, can be used in auditoriums, do notrequire installation of an antenna, and provide bettersound quality . It seems that AM systems will prob-ably disappear from the market in the near future.

Infrared light—The infrared-light listening systemstransmit audio signals by using an infrared lightbeam . The nignal from a microphone is fed into ahigh-frequency transmitter which drives the infraredemitters . The transmitter and light emitter are builtas one unit for small portable systems or as inde-pendent units for multi-emitter systems in largehalls . These systems differ from audio- or radio-frequency systems mainly in the frequency bandused for transmission . Infrared light transmissiontakes place in the frequency bands immediatelybelow the frequencies corresponding to visible light,i .e ., somewhere around 3 x 10" Hz.

The infrared systems available on the market arebased on the design of Sennheiser ElectronicCompany of Germany . The device used to producethe infrared light in the Sennheiser system is thegallium arsenide (GaAs) light-emitting diode (LED)which emits light beams with a wavelength of 930

nm . One GaAs diode is sufficient to cover an areaof 3-to-5 m 2 , according to Nebozenko writing in 1982

(8) . For good transmission quality in large audito-riums, multiple-diode arrays have to be assembled.

A system that uses direct amplitude-modulationof the diode current is susceptible to a variety ofinterferences from other light sources . To limit

interferences, the audio signal is first used tomodulate the frequency of a low-frequency "sub-oar[i8r" electric signal . The resulting "FM" signalis then used to modulate the amplitude of the

diode current . Because changes in frequency carrythe information, other light sources in the room areless able, or unable, to distort the signal.

The infrared receiver has a photodetector diodemounted behind a henn\aphorioa! lens which in-creases the amount of light intercepted and pre-sented to the sensitive area of the diode . Thepossibility of interferences from ambient light isreduced by an optical filter which is transparentonly to infrared light.

The receivers are available either as a stethoscope-type device placed under the chin of the listener oras a chest-type receiver . The first type is designedfor listeners who do not wear hearing aids . Thesecond type is designed to work in combinationwith hearing aids or with separate earphones . Theinduction neck loop is used for coupling with apersonal hearing aid switched to "T" position.

Infrared light has properties similar to those ofvisible light . Like visible light, infrared rays are com-pletely absorbed by a black velvet curtain, and theyare reflected directionally by shiny metal surfaces.Except for escape through windows and otheropenings, infrared light will remain confined to theroom of its source . (That feature has special signifi-cance in the professional theater where preventionof illegal recording might be important .)

Infrared transmission is highly directional . Inlarge auditoriums, light reflections should provideomnidirectional reception . Useful reflections areobtained from room surfaces with light reflectioncoefficients greater than 0 .5 (where more than halfof the light energy is reflected).

Bright sunlight and other light sources whichcontain infrared energy interfere with infrared trans-mission . Incandescent (tungsten) light producesthe most detrimental form of interference ; this lightcontains a great dool of infrared energy, especiallywhen it is dimmed (8) . Infrared systems are primarily

designed for indoor use . The intensity of infraredradiation of sunlight is less indoors than outdoors.On a bright day, the intensity of outdoor infraredradiation can reach such high levels that infraredtransmission becomes impossible . Although fluo-rescent light includes only small amounts of infraredenergy, it too may produce interference if its sourceis close to the IR receiver. Thus the performance ofan infrared system in a large auditorium is verydependent onarchitectural andinterior design con-ditions 8uohaoref!eCtionafr0rnUoundar y ou rfao9e

and amount of ambient light . Infrared systems,except for the small-room units, are permanentlyinstalled by specialists .

-

°

45


FIGURE 1Electrical frequency responses of the three listening systems:AL (full line), FM (dashed line), and IR (dotted line) .

\\

\\\

_\\

20

100 200 500 1K 2K 5K 10K 20KFrequency (Hz)

METHOD

Classroom

The experiment was performed in a classroomWith ovo!umno of 91 .2 nn « (7.G x 5.0 x 2.4 m). Theclassroom had a floor carpet and acoustic tiles onthe ceiling. Two large windows had single glasspanes and no drapes . The major source of back-ground noise in the classroom came from highwaytraffic . The average SPL of noise during day hourswas 46 dB (A) . The average reverberation time (T) ofthe room for the 500, 1000, and 2000 Hz octave bandsof noise was 0 .35 s, measured with a pink noise.The critical distance was calculated as equal to 1 .1meters.

Amplification Systems

Three listening systems, AL, FM, and IR, werecompared among themselves and with a PAsys-tem.

The PA system consisted of two identical loud-speakers (Bang and Olufson Beovox S45) located0.5 meters apart on the short wall of the classroom.

The transmitting AL system was specially de-signed for the classroom : see Letowski, Donahue,and NaUb!ek. 1985 (9). The system contained apower amplifier and equalizer used for smoothingfrequencies response.

The FM system was a commercially availablePhonic Ear PE 551A and PE 551T unit designed forauditorium use.

The \R transmitter was a commercially availableSiemens S1 406 unit advertised for home use . Allthree transmitters were suitable for use in amedium-size c!aaaroorn .

The following receivers were used for testingwithout hearing aids:

1. The receiver for AL was composed of a PhonicEar AT 119 neck loop, connected to a Radio ShackArcher miniamplifer with volume control and astethoscope;

2. The FM receiver was a Phonic Ear PE 555Rwith a stethoscope, and

3. The !R receiver was a Sennheiser HDI 407-8with a stethoscope . The same Phonic Ear AT 360stethoscope, composed of a button-type transducerand tubes, was used on all three receivers.

The hearing aid wearers used their own monauralhearing aids with a microphone input when testedusing the PA system, and they used a telecoil input(the aidinputevv!tohinthe°T"Ponition)vvhentemtedusing the three listening systems . The hearing aidwas the only receiver these subjects needed for theAL system . For the FM and IA systems, the respec-tive receivers with a Phonic Ear AT 119 neck loopwere coupled to the hearing aid.

The electrical frequency responses of threelistening systems are shown in Figure 1 . Thesewere obtained in the test room between the input towhich the signal-source tape recorder would beconnected and the output to which the stethoscopewould be connected . All three systems had flatwideband frequency response.

For all electroacoustic measurements, the volumecontrol of the listening system receivers was set tothe middle position of the gain control . Due to thetransmission levels, all subjects used receivervolume control settings of no greater than mediumgain . This \ovel setting was deemed more represen-tative of the output received by our subjects than

46


f

FIGURE 2Electroacoustic frequency responses of thethree systems at the output of the Phonic EarAT 360 button-type transducer (as in Figure 1).

Frequency (KHz)

FIGURE 3Electroacoustic frequency responses of thethree systems at the output of the Phonic EarAT 360 stethoscope tube (as in Figure 1).

.12 .25 .5

1

2

4

Frequency (KHz)

FIGURE 4Electroacoustic frequency responses of theFM system, Phonic Ear PE 551A amplifier,PE 551T transmitter, and PE 555R receiver atthe output of the Phonic Ear 841 LA hearingaid ; hearing aid telecoil input coupled througha Phonic Ear AT 119 neck loop to the FMreceiver . With perpendicular positioning ofthe loop and telecoil (full line) ; with parallelpositioning (dashed line).

Frequency (KHz)

12 25 .5 1

2 4

8

Frequency (KHz)

FIGURE 5Electroacoustic frequency responses of theIR system, Siemens SI 406 transmitter andSennheiser HDI 407-S receiver, with thehearing aid as in Figure 5 .

47


having the volume control set to full-on gain for thee!entromoouationn*aaun*nnente.

The electroacoustic frequency response of eachsystem is modified by an electroacoustic transducerand the manner in which acoustic energy is deliveredto the ear . Frequency responses of some arrange-ments were measured using a FONIX 5500Z HearingAid Test Set . The output of the FONIX Test Set wasfed into the power amplifier of the room loop or tothe line inputs of the FM and IR transmitters . Thestethoscope of the system receiver was placed inthe sound chamber of the FONIX Test Set . Thetubing of the stethoscope was coupled to the 2cccoupler and sealed with putty . Figure 2 shows a!oo-troacounticfrequoncy responses of the three sys-tems at the output of the Phonic EarAT30O button-type transducer . It can be seen that the responsesremained reasonably flat, with some high-frequencyattenuation . Subsequent modification of the fre-quency responses was caused by the stethoscopetubes (Fig . 3); the long tubes introduced largeirregularities and an even further limitation of high-frequency energy. Average harmonic distortions forvarious configuration of the systems for 500, 800and 1600 Hz ; respectively, were as follows : 0, 1, and1 percent for AL ; 4, 1, and 1 percent for FM ; and 2, 1,and 1 percent for IR.

The hearing aid users had their hearing aidte!e'coil inputs coupled with the receivers by means ofa neck loop . Since the subjects used various hearingaids, there was no common frequency response forthat listening mode . As an 'example, frequencyresponses of the FM and IR systems with the respec-

five receivers coupled'through the Phonic Ear neckloop AT 119 to a Phonic Ear hearing aid 841 LAtelecoil are shown in Figures 4 and 5, respectively.Because there was no uniform setting of hearingaid volume control among the subjects, the hearingaid volume oontrol having number indicators ofzero to 4 was set at 2 .5. For both receivers, theresponses depended on the relative position of theneck loop and the telecoil . When the loop and tele-coil vvereparpondiou!or. maximum SPL at 900 Hzwas greater than with the parallel arrangement by3.5 dB and 6.0 dB for the FM and IR receivers,respectively.

Subjects

Four groups of subjects were tested:1. Young adults, mean age 23 .5, range 20 to 25

years, with audiometrically normal hearing (hearingthreshold !avel no greater than 10 dB at any of theaudiometric frequencies between 250 Hz and 8000Hz) .

2. Hearing impaired with mild-to-moderate hear-ing loss not using hearing aido, mean age 62.8 'range 44 to 72 years.

3. Hearing aid users with motlerate hearing loss,mean age 54.2, range 22 to 74 years.

4. Elderly, mean age 71 .7, range 66 to 77 years,with hearing losses typically expected for their age.

There were 10 subjects in each group except forthe last group which consisted of 9 subjects . Theaverage air conduction thresholds using criteriadescribed by ANSI, 1976 (10) for the last three groupsare shown in Figure S.

-0 Right Ear

FIGURE 6Hearing threshold lin decibels for threegroups of subjects.

-

20 _

-

-

~ ElderlyD Hearing-Impairedo Hearing Aid User

250 500 1K 2K 4K 8K

250 500 1K 2K 4K 8K

Frequency (Hz)

_

48

NABLEK et al . : Amplification Systems in a Classroom

Procedures

The Modified Rhyme Test published by Kreul et al.in 1968 (11) was played from a tape recorder . Thesame recordings were previously used by Nabelekand Robinson, 1982 (12) . The output of the taperecorder was connected to the inputs of the PAsystem and all three listening systems . In the casesof the PA and AL systems, the tape recorder wasconnected to the respective amplifiers . For the FMand IR systems, the line input was utilized . Theseinputs are designed for use with tape recorders.The microphone input was not tested in this studybecause the main goal was to compare the systems.The systems had different input impedances andwere designed to be used with different micro-phones; if different microphones had been used,comparisons between systems wOuld have beendifficult.

The PA system was constantly activated duringall testing . The other systems were activated onlywhile they themselves were being tested . Thisarrangement simulated the typical situation in anauditorium where a listening system is used as anauxiliary amplification for selected listeners at thesame time that the PA system is used by the generalaudience.

The speech level from the PA system was set at66 dB (A) . This !evel was measured at 3 meters infront of the loudspeakers with a Rion Sound LevelMeter NA-21 . As a calibration signal, a segment ofpink noise recorded at the beginning of each speechtest (at a level corresponding to average peak levelsin test words) was used. This level was kept constantthroughout the experiment.

There were two modes of presenting the MRT:with and without a 12-voice babble . Babble levelwas set at 58 dB (A) as measured at the same pointas the speech level measurement . The babble wasreproduced by a third loudspeaker placed midwaybetween the two loudspeakers producing the MRT.The subjects were seated 3 meters in front of theloudspeakers. The room SIN at this distance was+ 20 and +8 dB, without and with masking noise,respectively . The first (+2O)S/N represents listeningconditions which are adequate for all subjects whilethe second (+ 8) SIN represents listening conditionswhich tend to cause pern8ptual problems for hear-ing impaired and elderly listeners . Lower SIN's

would be even more difficult but they are atypicalof classroom conditions. The reverberation time of

0 .35 s, typical for acoustically treated classrooms,tends to be a source of perceptual errors for hearingimpaired listeners: see Nabe!okand Pickett, 1974

( 2 ), and for elderly listeners : see N~bve!ek and

Robinson, 1982 (12 ) .Listening modes depended on whether or not the

subject used a hearing aid, and on the system beingtested . Listeners without hearing aids were testedin the following ways:

1. PA system with unaided ears, therefore therewas no control of the listening level;

2. AL system with a neck loop, amplifier withvolume control, and a stethoscope, with the oppor-tunity to adjust the loudness to a comfortable level;and

3. FM system with an FM receiver, a stethoscope,and the opportunity to adjust the loudness with areceiver volume control ; and

4. !R system with an !R receiver with a ntetho'aoope. and opportunity to adjust the loudness witha receiver volume control . The stethoscope wasdelivering acoustic signals to both ears.

The hearing aid users had binaural hearing lossesbut were using only one hearing aid . These subjectswere tested using their own hearing aids . The hear-ing aids, of different makes, all had telecoil (T)switches. The hearing aid users were allowed toadjust the hearing aid gain control only two timesduring each session : first, for the microphone inputwhile listening to the MRT without babble throughthe PA system, and later (for the te!ncoil input) whilelistening to the MRT through the AL system . Duringthat condition, the PA was also activated and therewas no babble masking . This to!eoo\l gain settingwas then left unchanged for listening to the FM andIR systems. (If desired, the subjects could use thevolume control of the respective receivers, but notof their aids, to additionally adjust the loudnesslevel .)

Each condition was tested with two MRT listsequalling a total of 100 words. All eight conditions(four systems, each tested with and without babblenoise) were tested using different randomizationsof the MRT word lists.

The conditions were tested in the following order:PA without babble, PA with babble and AL withoutbabble. The remaining conditions, FM and IR withand without babble in the room, and AL with babble,were tested in random order. The AL without babblewas tested before FM and IR systems so that thehearing aid users could adjust their gain controlsfor the telecoil input.

RESULTS

The word-identification scores in percentage cor-rect were transformed to the proportions p, whichin turn were converted to arcsins \FP, using the

49


rAaLs1Mean word-identification scores and standard deviations (sd) in percentage onneo, trans-formed from mean 4) scores for two room speech-to-noise ratios, four systems, and four groups

of subjects.

Group

SYSTEM

PA AL

FM IR

~

mean sd mean sd

mean

(SIN = +20dB)

Normal 091 0 .5 09 .9 0 .4

89 .9 0 .2 99 .8 0 .4

Elderly 90 .5 0 .2 96 .9 0 .4

98 .4 0 .4 98 .2 0 .4

Hearing impaired 89 .0 1 .5 95 .2 1 .2

95 .8 1 .3 93 .0 2 .1

Hearing aid users 86.9 1 .2 92 .2 1 .6

92 .0

(SIN = +8dB)

2.5 89 .6 2 .3

Normal 93 .6 0 .7 90 .9 0 .3

90 .9 0 .3 99 .9 0.4

Elderly 94 .3 0 .4 97 .0 1 .0

98 .5 0 .7 98 .0 0 .5

Hearing impaired 79 .9 1 .2 96 .2 2 .0

97 .0 1 .5 93 .7 1 .5

Hearing aid users 83 .1 1 .1 90 .1 2 .8

91 .1 1 .6 87 .3 1 .5

method of Walker and Lev, 1053(13)to stabilize theerror variance . The following equation was used:

~aroainvr~

The 0 values served as the criterion measure.Analysis of variance was performed using a re-

peated measure model . A three-factor design wasused to analyze the effect of group (G), system (S),and room signal-to-noise ratio (SIN) . All main effectsand two two-way interactions were statisticallysignificant at p < 0 .001:

G [F(3,35) = 385.47],

S [F(3 `105) = 105 .76],

SIN [F(1,35) = 15 .12],

G x S [F(9,105) = 6 .37], and

S X 8/N (F(3 ' 105) = 16.84].

The interactions G XSIN and G GxSx SIN werenot statistically significant . The mean word identi-fication scores obtained from mean 0 scores byreversed transformation are in Table 1 and in Figure7. Consequently, all reported means were obtainedby the transformation from 0 scores.

The effect of groups reflected different degreesof hearing loss . The mean scores were as follows:99.6, 97 .5, 93 .1, and 89 .2 percent for normal-hearing,elderly, hearing impaired and hearing aid groups,

respectively. Duncan's multiple-range test indicatedthat all four means were different.

The means for the four systems were as follows:97.5.97.0,96.2 and91 .3percent forFK4 ' AL. !RandPA systems, respectively . Duncan's multiple-rangetest indicated that the two first means (for the FMand AL systems) were not different, while the othercontrasts were significantly different.

The effect of room SIN was significant but thedifference between the means was small (96.3 per-cent and 95 .3 percent for f20 dB and +8 dB,respectively).

The two- way interaction G x S indicated thatthe system performance depended upon the group.Post hoc least-squares means analysis was per-formed to determine the relationship between thesystems and groups . Fora!! four groups, the PAsystem was significantly different from the remain-ing three systems (p < 0 .005) . Furthermore, therewas no significant difference among the AL, FM,and IR systems for normal hearing and elderly sub-jects . For the hearing impaired subjects and hearingaid users there was a significant difference betweenthe AL, FM, and IR systems. For the first group, theIR was different from both the AL and FM but theAL and FM were not different . For the hearing aidgroup, the AL was not significantly different fromthe FM and lR but the FM was different from lR.

50


FIGURE 7Word identification scores for four amplification systems, two speech-to-noise ratios and four groups of subjects.

S/N = +20 dB

S/N =+ B dB

80 * Normal-Hearing• Elderlyq Hearing-Impairedo Hearing Aid Users

Or I I1 T

PA AL

FM

IR

PA AL FM

IR

System

0 100

U)

0

0 90

4-Cq ~WV

0

•

The two-way interaction S x SIN indicated thatthe system performance depended on the SIN . Thedifferences between the two S/N's for the AL, FM,and IR systems were less than 1 percent while thedifference for the PA system was 6 percent . Toillustrate more clearly that the performance of onlythe PA system depends on the SIN, an additionalgroup of 10 normal hearing subjects was tested ata S/N of 0 dB with all four systems, and with the PAat SIN of + 20 dB. The performance of this secondgroup at S/N of + 20 dB was comparable to the per-formance of the first group (98 .7 percent and 99 .1percent respectively) . The results of the first groupat S/N of + 20 and + 8 dB, and the second group atSIN of 0 dB, are shown in Figure 8 .

FIGURE 8Word identification scores as functions of speech-to-noise ratiofor four amplification systems with normal hearing subjects .

p

51


DISCUSSION

The results indicate that speech perception withany of the three listening systems was better thanwith the PA system.

For normal-hearing and elderly subjects, therewas no significant difference in performance of theAL, FM, and IR systems.

For the hearing-impaired listeners, the perfor-mance of the !R system (93 .496) was worse thanthat of the AL (95 .7%) or of the FM (96 .4%) system.While the differences were statistically significant,they were small and should not be consideredimportant.

For hearing aid users, the performance with theAL (912Y6) was not significantly different fromthe performance with the FM (91 .8Y6) or the \R(88.5%) systems, but the performance with the FMwas better than performance with the IR system.Here again, while the difference was statisticallysignificant, it was small.

It can be concluded that all three listening systemswere equally effective inovercoming acoustic degra-dations. Theimprovement, incomparison with per-formance when relying on the PA system, wasmodest, but for those tests the listening conditionsin the room were relatively good ; an additionalexperiment with normal hearing subjects indicatedthat when room noise was increased, the relativeimprovement was greater.

The good performance of the AL system was prob-ably by its proper design and use of theequalizer. The equalizer was responsible for thesimilarity of the AL frequency response to the fre-quency responses of the FM and IR systems . Withoutthe equalizer, the frequency response of the AL sys-tem was irregular and narrower, as reported byLetovvaki. Donahue, and Nabe!ek in 1985(8) . How-ever, even when tested without the equalizer, thefrequency response of this audio loop was muchbetter than that reported in 1970 by Sung, Su n g ,Hodgson, and Angelelli, who found a very irregularand narrow frequency response in a study of a class-room audio induction loop . Also, the special designof the loop for this study assured high homogeneityof field strength . The results of the speech percep-tion testing indicate that a well-designed audio loopcan be equally as effective as an FM or IR system.

Effects Introduced by Transducers

The frequency responses of all three listeningsystems were greatly modified by electroacoustictransducers. The frequency responses became ir-regular and high-frequency energy was attenuated .

For subjects listening without hearing aids, largeirregularities were introduced by the stethoscopetubes . Similar irregularities were reported by Haw-kins ' Fluck, and Van Meter in 1982 (4) for the 442PPhonic Ear FM receiver with a stethoscope, measuredthrough aKEK4ARanthnnponnetrioacoustic manikinwith Zwislocki ear simulator.

For subjects listening with hearing aids, the fre-quency response was dependent on the couplingbetween the neck loop and the ta!eco\l in the hear-ing aid . We had found (in a pilot study) that thefrequency responses of neck loops vary ; therefore,one neck loop was used throughout the study . Wehad no control over the telecoils because varioushearing aids were used by the subjects . On a selectedhearing aid it was shown that the transmission oflow frequencies, and overall gain, depended on therelative position of the loops arid the coil . T|h"e gtg@in was achieved with the two p

iito each other, while the lowest gain was achievedwith them parallel to each other . The same depen-dance on position was found by Hawkins and VanTasell, who reported in 1982 that output differenceswere 1 to 10 dB for several hearing aids coupled viathe neck loop to an FM receiver (15) . Consequently,when hearing aids are coupled via the neck loop toa system receiver, the overall gain and frequencyresponse can vary with head and body movements.This could have influenced the performance of thehearing aid users with the FM and IR systems.

A possible explanation for the relatively lowerscores with the \R system is that one of the two IRreceivers that were used produced oonaaional in-ternal noise. This was not discovered until the endof the project . Another possible influence ieoug-geated by the fact that all tests were performedduring daylight hours . However, any interferencecaused by excessive light, if it occurred, failed tOaffect the scores of either the normal hearing or theelderly subjects.

The hearing-impaired subjects without hearingaids performed better than hearing aid users in allconditions -- except with PA at SIN of f8 dB . Theotherwise superior performance of the group wasrelated to its relatively better hearing thresholdlevels. However, with the PA, the subjects withouthearing aids had no control over the listening level,which might have been too low, and obtained scoreseboui3 percent lower than the hearing aid users.Such a situation is quite typical in real life, wherethe listening levels are set for nornnal hearing lis-teners. The amplification systems for hearing im-paired listeners allow individual control of listeninglevel.

.

52


The sound can be delivered to the ears in arrange-ments other than those tested . The tested arrange-ments were probably the worst for distortion of thefrequency response . This becomes apparent fromcomparison of Figures 2 and 3 . Bankoski and Ross in1984 (7), reported on tests of two FM receivers, onewith the transducer placed in headphones andanother with the transducer placed at the entranceof the stethoscope tubes . The difference for thetwo arrangements was not significant ; however,there was a trend in scores in favor of the head-phones arrangement . For the hearing aid users, acoupling superior to the neck-loop-and-telecoilwould be a direct connection between the FM outputand the hearing aid input (15) . Hawkins and VanTasell reported in 1982 that, with such direct con-nection, the frequency response of the receiverwith the hearing aid becomes very similar to thefrequency response of the hearing aid with a micro-phone input (15) . Therefore the use of headphonesor insert earphones and direct connections to thehearing aids should be preferable.

The systems in this study were compared in amedium-size room because installation of an ALand IR system in a large auditorium would be muchmore expensive . However, the results obtained inthis study can be applied to space of any size, pro-viding that the system will secure adequately strongand homogeneous transmission.

All three tested listening systems seem to besuitable for listeners with various degrees of hearinglosses. They can, be used by listeners with and with-out hearing aids. The decision about selection ofone system should be based on considerationsother than speech perception, such as cost, installa-tion requirements, maintenance, interferences, andmobility

Acknowledgements

The authors gratefully acknowledge the collabora-tion of Scott Posner and Phonic Ear Corporation forthe loan of a Phonic Ear FM listening system andaccessories .

Y 4

REFERENCES

1. Knudsen VO : The hearing of speechin auditoriums. J Acoust Soc Am1 :56-82, 1929.

2. Nabelek AK, Pickett JM : Monauraland binaural speech perceptionthrough hearing aids under noise andreverberation with normal and hear-ing impaired listeners . J Speech HearRes 17 :724-739, 1974.

3. Finitzo-Hieber T, Tillman TW: Roomacoustics effects on monosyllabicword discrimination ability for normaland hearing impaired children . JSpeech Hear Res 21 :440-458, 1978.

4. Hawkins DB, Fluck JF, Van MeterSL: Implementation of an amplifica-tion system to assist the hearingimpaired in a university setting.ASHA 24(4) :263-267, 1982.

5. Vaughn GR : Assistive listening de-vices . Part 2 : Large area sound sys-tems . ASHA 25(3) :25-30, 1983.

6. Williams GI : Hearing assistance sys-tems technology . Sound & VideoContractor, pages 12-19, January,1984.

7. Bankoski, SM, Ross M : FM systems'effect on speech discrimination inan auditorium. Hearing Instruments35(7) :8-12, 1984.

8. Nebozenko J : Infrared Listening Sys-tems in the Theatre . New York : AudioEngineering Society, Preprint No.1926, pages 1-7, 1982.

9. Letowski T, Donahue AM, NabelekAK: An induction loop system de-signed for a classroom . J Rehabil R D22(2):63-69, 1985.

10. Specifications for Audiometers(ANSI S1 .4) . New York, AmericanNational Standards Institute, 1976.

11. Kreul EJ, Nixon NC, Kryter KD, BellDW, Land JS, Schubert EG : A pro-posed clinical test of speech dis-crimination . J Speech Hear Res11 :536-552, 1968.

12. Nabelek AK, Robinson PK : Monauraland binaural speech perception inreverberation for listeners of variousages . J Acoust Soc Am 71 :1242-1248,1982.

13. Walker HM, Lev J : Statistical Inter-ference . New York : Holt, Rinehart &Winston, 1953.

14. Sung RJ, Sung GS, Hodgson WR,Angelelli RM : Performance of hearingaids with an induction loop amplifica-tion system : laboratory versus class-room setting. Audiol 15 :249-256,1976.

15. Hawkins DB, Van Tasell DJ : Electro-acoustic characteristics of personalFM systems . J Speech Hear Disord47 :355-362, 1982 .

tvi‘., veterans administrationanna k. nabelek, ph . d. amy m. donahue, m.a. tomasz r. letowski,...

Documents