effect of multichannel digital signal processing on

Washington University School of Medicine Washington University School of Medicine

Digital CommonsBecker Digital CommonsBecker

Publications Division of Adult Audiology

2006

Effect of multichannel digital signal processing on loudness Effect of multichannel digital signal processing on loudness

comfort sentence recognition and sound quality comfort sentence recognition and sound quality

Karen M Mispagel Washington University School of Medicine in St Louis

Michael Valente Washington University School of Medicine in St Louis

Follow this and additional works at httpsdigitalcommonswustleduaudio_hapubs

Recommended Citation Recommended Citation Mispagel Karen M and Valente Michael Effect of multichannel digital signal processing on loudness comfort sentence recognition and sound quality (2006) Publications Paper 1 httpsdigitalcommonswustleduaudio_hapubs1

This Article is brought to you for free and open access by the Division of Adult Audiology at Digital CommonsBecker It has been accepted for inclusion in Publications by an authorized administrator of Digital CommonsBecker For more information please contact vanamwustledu

Delivered by Ingenta to Washington University School of Medicine LibraryIP 1282521042 On Tue 20 Sep 2011 184911 681

J Am Acad Audiol 17681ndash707 (2006)

Department of Otolaryngology-Head and Neck Surgery Division of Adult Audiology Washington University School ofMedicine St Louis Missouri

Karen Mispagel Washington University School of Medicine Department of Otolaryngology-Head and Neck SurgeryDivision of Adult Audiology 660 South Euclid Ave Campus Box 8115 St Louis MO 63110 Phone 314-362-7490 Fax 314-747-5593 E-mail mispagelkentwustledu

Effect of Multichannel Digital SignalProcessing on Loudness Comfort SentenceRecognition and Sound Quality

Karen M Mispagel

Michael Valente

Abstract

This study evaluated the effect of increasing the number of processing channelsfrom 32- to 64-signal processing channels on subjectsrsquo loudness comfort andsatisfaction sentence recognition and sound quality of his or her own voiceTen experienced hearing aid users with mild-to-moderate sensorineural hearingloss wore behind-the-ear (BTE) hearing aids with Adaptive Dynamic RangeOptimization (ADROtrade) signal processing for a period of six weeks in the 32-channel and 64-channel conditions Results revealed no significant differencesin loudness comfort or satisfaction for the majority of sound samples asmeasured by the Subjective Loudness Test and Environmental SoundsQuestionnaire No significant differences in sentence recognition between thetwo processing conditions were found as measured by the Hearing In NoiseTest (HINT) Additionally no subjective differences in sound quality of subjectsrsquoown voice were determined by the Listening Tasks Questionnaire

Key Words Adaptive dynamic range optimization (ADRO) Hearing In NoiseTest (HINT) multichannel compression R-Spacetrade noise

Abbreviations ADROtrade = Adaptive Dynamic Range Optimization ANSI =American National Standards Institute BTE = behind the ear CLL = comfortablelistening level DSP = digital signal processing HINT = Hearing In Noise TestIRB = institutional review board MIL = most intelligible level OB = octave bandPTAlf = low-frequency pure-tone average RTS = reception threshold forsentences SNR = signal to noise ratio

Sumario

Este estudio evaluoacute el efecto de incrementar de 32 a 64 el nuacutemero de canalesde procesamiento de la sentildeal sobre el nivel agradable de intensidad subjetivadel sujeto y la satisfaccioacuten el reconocimiento de frases y la calidad del sonidode su propia voz Diez sujetos con experiencia en el uso de auxiliares auditivoscon hipoacusias sensorineurales leves a moderadas utilizaron auxiliaresretroauriculares (BTE) con Optimizacioacuten Adaptativa del Rango Dinaacutemico(ADROtrade) para procesamiento de la sentildeal por un periodo de seis semanasen condiciones de 32 y 64 canales Los resultados no revelaron diferenciassignificativas en el nivel confortable de intensidad subjetiva o en la satisfaccioacutenpara la mayoriacutea de la muestra de sonidos medidos por medio de la Pruebade Intensidad Subjetiva y el Cuestionario de Sonidos Ambientales Tampocode encontraron diferencias significativas en el reconocimiento de frases entrelas dos condiciones de procesamiento medidas con la Prueba de Audicioacutenen Ruido (HINT) Ademaacutes no se determinaron diferencias subjetivas en la

Delivered by Ingenta to Washington University School of Medicine LibraryIP 1282521042 On Tue 20 Sep 2011 184911

Journal of the American Academy of AudiologyVolume 17 Number 10 2006

682

Digital signal processing (DSP)algorithms have become increasinglymore complex since DSP hearing aids

became widely commercially available in themid-1990s DSP hearing aids are nowavailable with as many as 32 channels ofsignal processing As the number ofprocessing channels has increased studieshave investigated the potential advantagesof multichannel signal processing (Mooreand Glasberg 1986 Kiessling and Steffens1991) Relative to single-channel compressionmultichannel processing can increaseintelligibility because of the increase in theaudibility multichannel compression providesfor low level input sounds (Dillon 2001)Additionally multichannel compressionallows the frequency response of hearing aidsto be more easily controlled by providingprogramming flexibility not available insingle-channel processing (Kuk 2002)Multichannel processing has also been criticalin the development of more effective noisesuppression and feedback managementstrategies With an increased number ofprocessing channels noise reduction andfeedback strategies can more preciselydecrease gain in channels in which noise orfeedback is occurring with less reduction inthe adjacent frequency channels or thosecontaining speech

Although advantages of multichannelsignal processing are documented potentialnegative side effects of increasing the numberof channels have been investigated (Crainand Yund 1995 Moore et al 1999 Stoneand Moore 1999 2002 2005 Agnew andThornton 2000) Possible disadvantages ofmultichannel signal processing includechannel summation temporal (or spectral)smearing and increased group delay

Channel summation occurs when theoutput in each channel of a multichannelhearing aid combines resulting in a widerbandwidth than any individual channel andan increase in overall output (Dillon 2001)In general the more channels present and thehigher the compression ratio in each channelthe greater the summation effect (Kuk andLudvigsen 2003) If unaccounted for thisincrease in output as a result of multichannelprocessing could lead to greater loudnessdiscomfort for the hearing aid user whencompared with hearing aids with fewerchannels Kuk and Ludvigsen (2003)demonstrated this by evaluating the outputof four hearing aids with different numbersof channels Results revealed that the outputof the 15-channel hearing aid was almost 10dB greater than the single-channel aid andat least 5 dB greater than the two- and three-channel hearing aids The results from thisinvestigation suggested that channelsummation may need to be accounted duringthe fitting of hearing aids Dillon (2001)recommended a reduction in real earsaturation response (RESR) levels formultichannel hearing aids in which theoutput is controlled independently in eachchannel The amount of recommendedreduction increased as the number ofchannels increased two channelsmdash5 dB reduction three channelsmdash7 dBreduction four channelsmdash9 dB reductionand five channelsmdash10 dB reduction Norecommendations were made for hearingaids containing more than five channels

As mentioned above another majorconcern of increasing the number of signalprocessing channels is temporal (or spectral)smearing Temporal smearing occurs whenthe intensity difference between the peaks

calidad de sonido de la propia voz del sujeto determinada por el Cuestionariode Tareas de Audicioacuten

Palabras Clave Optimizacioacuten adaptativa del rango dinaacutemico (ADRO) Pruebade Audicioacuten en Ruido (HINT) compresioacuten multicanal ruido R-Spacetrade

Abreviaturas ADROtrade = Optimizacioacuten adaptativa del rango dinaacutemico ANSI= Instituto Americano Nacional de Estaacutendares BTE = retro-auricular CLL =Nivel confortable de audicioacuten DSP = procesamiento digital de la sentildeal HINT= Prueba de Audicioacuten en Ruido IRB = comiteacute institucional de revisioacuten MIL =nivel maacutes inteligible OB = banda de octava PTAlf = promedio tonal puro abajas frecuencias RTS = umbrales de recepcioacuten para frases SNR = tasasentildealruido


Multichannel Signal ProcessingMispagel and Valente

683

and troughs in the speech envelope is reducedIn multichannel hearing aids as the numberof channels increases the intensity differencedecreases (Kuk 2002) This reduction intemporal contrasts could cause a reduction inspeech recognition especially for those withgreater than a moderate hearing loss who relyon temporal contrasts for speech recognition(Van Tassell et al 1987) Numerous studiesinvestigating the effect of the number ofchannels on speech recognition have beenundertaken and the results from thesestudies have been variable (Summerfield1992 Crain and Yund 1995 Moore et al1999 Stone and Moore 1999 2002 2005Agnew and Thornton 2000) Crain and Yund(1995) investigated the ldquodegradation of voweland stop-consonant discrimination as afunction of the number of channels andcompression ratiosrdquo (p 530) The results forhearing-impaired subjects indicated whenthe multichannel compression processingstrategy was customized to the subjectsrsquohearing loss utilizing subject specificthreshold and loudness discomfortinformation no significant change in voweldiscrimination performance wasdemonstrated as a result of increasing thenumber of channels Additionally it wasreported that significant discrimination errorswith vowel spectra were present only whencompression ratios were high (value notreported by the authors) in each channel andwhen the number of channels was greaterthan eight

Moore et al (1999) evaluated theeffectiveness of multichannel compressionwith one two four and eight channels bymeasuring subjectsrsquo speech receptionthreshold (SRT) with the Hearing In NoiseTest (HINT) sentences Although only a slightbenefit of multichannel compression wasseen in this study the authors theorized thatldquofurther increases in the number ofcompression channels with correspondingreductions in bandwidth of each channelmight lead to a system that was more effectivein improving the detectability of portions ofthe speech target falling in the spectral dipsin background soundrdquo (p 409)

Yund and Buckles (1995) investigatedthe effect of increased number of processingchannels on speech recognition of mild-to-moderately severe hearing-impaired subjectsReported results indicated a highly significanteffect for number of channels Increasing

from four to eight processing channelsimproved speech recognition but above eightchannels no further improvement was found

The final potential disadvantage ofincreasing the number of signal processingchannels is group delay ldquoProcessing timerdquo orldquogroup delayrdquo is defined as the finite timedelay created as an input signal passesthrough a hearing aid from the microphoneto the receiver (Agnew and Thornton 2000)The group delay in digital hearing aids isconsiderably longer in comparison to analoghearing aids due to the complex conversionof the input sound signal into discretequantities for signal processing Whereasthe time required for analog hearing aids toprocess input signals is very short a fewtenths of a millisecond (msec) the timeneeded for DSP can vary widely dependingon the DSP algorithm In general as theamount of processing increases so does theprocessing time or group delay (Frye 2001)

Previous research has demonstrated thatlong group delay can negatively affect speechproduction and perception for normal-hearingand hearing-impaired patients (Summerfield1992 Stone and Moore 1999 2002 2005Agnew and Thornton 2000) Specificallyconcerns of auditory confusion (Summerfield1992) and degradation of speech productionand perception of subjectsrsquo own voice (Stoneand Moore 1999 2002 2005 Agnew andThornton 2000) as a result of delay havebeen investigated

Auditory confusion can occur when thereis a delay between the hearing aid userobserving the movement of the talkerrsquos lipsand hearing the sound of his or her voiceSummerfield (1992) reported that sound canlag the visual image by more than 80 msecbefore confusion will occur Therefore herecommended that processing for hearingaid users with severe-to-profound hearingloss be as short as possible but group delaysas long as 40 msec would be acceptable

Stone and Moore (1999) reported on theeffect of delay on a subjectrsquos own speechproduction and perception of his or her ownvoice for normal-hearing populations using asimulation of hearing loss They reportedthat delays greater than 20 msec can lead tothe perception of an ldquoechordquo in the subjectsrsquo ownvoice whereas delays less than 10 msec mightlead to a perception of a subtle change in thetimbre of the sound In a follow-up studyStone and Moore (2005) utilized hearing-



684

impaired subjects to measure the effect ofgroup delays (13ndash40 msec) on perception ofthe subjectrsquos own voice and speech productionIt was concluded that subject disturbance tothe sound of his or her voice increased withincreasing group delay Additionally subjectswith low-frequency (500 1000 and 2000 Hz)hearing loss greater than 50 dB HL weresignificantly less disturbed than thosesubjects with less low-frequency hearing lossSpecifically the results showed that delaysgreater than 15 msec can be unacceptable tolisteners with low-frequency hearing lossaround 35 dB HL but those with moremoderate-to-severe hearing loss or very mildhearing losses in the low frequencies maybe able to tolerate longer delays

Stone and Moore (2002) analyzedobjective and subjective measures of effectsof hearing aid delay on speech productionand perception in two different environmentswith the goal of defining an upper limit topermissible processing delay They concludedthat normal-hearing subjects reported thatdisturbing effects on perception becomesignificant when delays exceeded 15 msec inan office environment and 20 msec in a testbooth Objective measures of speechproduction did not show any significantnegative effects of delay until the delayreached 30 msec As a result of these findingsStone and Moore (2002) recommended DSPhearing aids which should be able toincorporate delays as long as 15 msec withfew negative side effects Additionally the amount of tolerable processing delay increased by 4 msec in reverberantenvironments compared to a near anechoicenvironment

Agnew and Thornton (2000) investigatedthe amounts of delay that were just noticeableand considered objectionable with 18 normal-hearing engineers to determine a worse caselimit for DSP hearing aid design Thelisteners in this study reported that timedelays greater than 10 msec wereobjectionable 90 of the time a significantlyshorter time delay than what was publishedby Stone and Moore (1999 2005)

Overall results from past research thatexamined the effect of increasing the numberof channels in DSP hearing aids suggest thatthe issues of channel summation temporalsmearing and group delay need to beaddressed when fitting multichannel hearingaids These effects have been well researched

in other compression strategies yet howthese issues affect loudness comfort andsatisfaction sentence recognition and soundquality of a hearing aid userrsquos own voice hasnot been addressed in Adaptive DynamicRange Optimization (ADROtrade) signalprocessing

The current study utilizes AdaptiveDynamic Range Optimization (ADROtrade)signal processing in 32- and 64-channelprocessing strategies ADROtrade is a slowlyadapting DSP that controls the output levelof a set of narrow-frequency bands so that thelevels fall within a specified dynamic rangeThe 32-channel processing strategy has a250 Hz bandwidth for each channel from 125to 8000 Hz The 64-channel processingstrategy has a bandwidth of 125 Hz for eachchannel By using narrow channels in thehearing aid there is great flexibility to shapethe maximum gain maximum output levelscomfort targets and audibility targets ineach channel allowing ADROtrade to be fit to awide range of hearing losses

The dynamic range of ADROtradeprocessing is defined by the threshold ofaudibility and a comfortable level withineach frequency channel for an individualADROtrade measures the peaks and troughs ofthe output signal unlike most amplifiers thatmeasure the average level of the input signalSince ADROtrade does not make an assumptionregarding the input signal dynamic range itcan maintain comfort and audibility of a widevariety of sounds not just speech (Blamey2005) A set of rules is implemented to controlthe output levels with the goal of keeping theoutput signal level within the optimumdynamic range First the ldquocomfort rulerdquorequires 90 of the output levels to be belowthe comfort target level in each frequencychannel This rule ensures that sounds arenot too loud Next the ldquoaudibility rulerdquorequires 70 of the output levels to be abovethe audibility target in each channel Thisensures that sounds are not too soft Theldquoaudibility rulerdquo is applied only if the ldquocomfortrulerdquo is satisfied The magnitude ofapproximate increase or decrease in gain orldquoslew raterdquo can be changed from the defaultof 3 dBsec to 6 dBsec through themanufacturer rsquos software The ldquohearingprotection rulerdquo limits the output level ineach channel so that it never exceeds themaximum output level Finally theldquobackground noise rulerdquo limits the maximum



685

gain in each channel to ensure low-levelbackground noise is not overamplified(Blamey 2005) As a result of the ADROrules gain does not change unless the comfortor audibility rule is violated

The studies described previously all haveexamined the impact of increasing the numberof processing channels in other compressionstrategies but it is unknown if similar resultswould occur with ADROtrade processing Martinet al (2001) compared ADROtrade signalprocessing to a linear fit hearing aid utilizingopen-set sentences at multiple intensity levelsSubject performance with ADROtrade processingwas significantly better than the linearprocessing at 55 and 65 dB SPL (159improvement at 55 dB and 36 dBimprovement at 65 dB) In a reverse-blockdesign study by Blamey et al (2004) thedifference in sentence recognition in quiet andnoise between a nine-channel-wide dynamicrange compression (WDRC) and a 64-channelADROtrade signal processing strategy wasexamined The results revealed that thesubjectsrsquo mean performance with ADROtradeprocessing was statistically significantly betterin quiet and noise than WDRC processingalthough it is unclear if the slight improvement(785 word score and 641 phoneme scorein quiet and 725 in noise) in performance isa result of increasing the number of processingchannels of difference in fitting strategy or ofdifferences in amplification strategies

The current study utilized a hearing aidwith ADROtrade processing in 32- and 64-channel strategies programmed using thesame in situ fitting method This comparisonbetween 32- and 64-channel ADROtradeprocessing examined whether detrimentalside effects (ie channel summationtemporal smearing or group delay) as resultof increasing the number of processingchannels occurred that could lead todecreased loudness comfort or satisfactionpoorer sentence recognition or decreasedsound quality of the subjects own voiceFurthermore if laboratory benefits ofincreasing the number of ADROtrade processingchannels are present these same benefitsshould ideally be accompanied by increasedreal-world benefits (ie external validity) inorder to establish the effectiveness of theprocessing strategy If on the other handlaboratory benefits of increasing the numberof ADROtrade processing channels are notpresent then it can be assumed that

increasing the number of channels would notprovide any significant benefit andperformance with a 32-channel processorwould provide the same level of performanceas a 64-channel processor

The primary objectives of the presentstudy were to determine if

1 Significant differences in loudnesscomfort were present between 32-and 64-channel processing strategiesas measured by the SubjectiveLoudness Test in the aided conditionand the Environmental SoundsQuestionnaire in the unaided andaided conditions

2 Significant differences in satisfactionwere present between 32- and 64-channel processing strategies asmeasured by the Subjective LoudnessTest in the aided condition and theEnvironmental Sounds Questionnairein the unaided and aided conditions

3 Significant differences were presentbetween 32- and 64-channelprocessing strategies in an adaptivedirectional microphone mode for thereception threshold for sentences(RTS in dB) required for 50performance on the Hearing In NoiseTest (HINT) sentences presented at0deg and diffuse R-Spacetrade noise (eightloudspeaker array) fixed at 65 dBA

4 Significant differences were presentbetween 32- and 64-channelprocessing strategies in the adaptivedirectional microphone mode forreception threshold for sentences(RTS in dB) required for 50performance on the Hearing In NoiseTest (HINT) sentences presented at0deg in quiet

5 Subjective differences in soundquality of subjectsrsquo own voice qualitywere present between the 32- and 64-channel processing strategies asmeasured by the Listening TasksQuestionnaire

PROCEDURES

Subjects

Ten adults (8 males 2 females meanage = 708 years SD = 114 years) with mild-



686

to-moderately severe bilateral symmetricalsensorineural hearing loss (ANSI [AmericanNational Standards Institute] 1996)participated in this investigation Symmetrywas defined as no greater than a 15 dB HLdifference in interaural thresholds at250ndash4000 Hz The magnitude of hearing losswas within the recommended fitting range forthe experimental behind-the-ear (BTE)hearing aids Pure-tone thresholds andacoustic immittance were measured duringthe first test session All subjects exhibitedno significant air-bone gap at any frequencyand normal tympanograms Mean wordrecognition scores under earphones in quietat the most intelligible level (MIL) were756 (SD = 107) and 774 (SD = 131)for the right and left ears respectively Thepresentation level to assess word recognitionat MIL is determined by monitored live voicepresentation (voice peaking at 0 on the VU[volume units] meter) of conversationalspeech and asking the subject to indicatewhen the presentation level was comfortablyloud and most intelligible

Figure 1 illustrates the mean hearingthresholds (average of right and left ears) at250 to 8000 Hz All subjects had priorexperience with binaural digital adaptivedirectional amplification for at least one yearwith their current hearing aids See Table 1for subject specific hearing aid information

The subjects were recruited from theWashington University School of MedicineAdult Division of Audiology When subjectswere recruited for the study they were askedto sign the institutional review board (IRB)approved consent form Subjects were told thepurpose of the study was to evaluate twodifferent processing strategies but they werenot informed about the signal processing orany other aspect of the experimental hearingaid Finally to compensate the subjects for hisor her efforts subjects were provided $200 atthe conclusion of the study

Fitting the Experimental Hearing Aids

ADROtrade processing in 32- and 64-channel amplification strategies was placed

Figure 1 Mean hearing thresholds averaged for the right and left ears Also provided is plusmn1 SD



687

into a BTE hearing aid supplied by one of thesponsors The aid had a volume control wheeland push button to access multiple programsA software program was utilized to alternatebetween the two signal processing strategiesthroughout the study After changingstrategies coupler and real ear measuresusing the Frye 6500CX were performed toensure accurate hearing aid performanceAdditionally the group delay was measuredon the Frye 6500CX test box Themeasurement was performed by using abroadband impulse signal and a 20 msectime window for each amplification strategy(Frye 2001) The mean group delay for the32-channel processing was 69 msec Themean group delay for the 64-channelprocessing strategy was 128 msec

The experimental hearing aids wereinitially fit using the manufacturer rsquos(Dynamic Hearing LTD) recommendedfitting procedure Briefly the hearing aidswere coupled to NOAH utilizing a QuickCOM(manufactured by AVR Communications LTD)interface box and the aids placed in the earcanal The QuickCOM box is a manufacturerspecific interface box that allows for fastercommunication between NOAH software andhearing aids The subjectrsquos own earmoldswere used to ensure a comfortable fit SeeTable 1 for subject earmold information Theresults from the audiometric test were usedto predict the initial comfortable levels atseven frequencies spaced at half-octaveintervals from 500 Hz to 4000 Hz Then theindividual in-situ comfortable listening levels(CLL) were measured using sixth-octavebands of noise in a bracketing procedure (2dB and 4 dB step sizes for up and downrespectively) To measure CLL a seven-pointcategorical loudness scale was used The

seven categories were (1) very soft (2) soft(3) comfortable but slightly soft (4)comfortable (5) comfortable but slightly loud(6) loud but OK and (7) uncomfortably loudSubjects were instructed to assign a loudnesscategory when each stimulus was presentedThe 50 intensity level that the subjectjudged to be at the highest level ofldquocomfortablerdquo on the loudness scale wasselected as the comfort target for each of theseven channels Then the noise was presentedin a sweep across frequencies to ensurebalance in comfortable loudness judgmentsacross frequencies If necessary the comforttarget was adjusted until the subject judgedthe stimuli to be equal in loudness acrossthe frequency channels Each hearing aidwas programmed with an omnidirectionalmicrophone in Program 1 adaptivedirectional microphone in Program 2 andtelecoil in Program 3 The volume controlwas programmed to provide a 20 dB SPLrange (10 dB SPL up and 10 dB SPL down)Finally fine-tuning adjustments were madeto the initial frequency response to addressany concerns related to feedback and theocclusion effect

After the initial fit subjects wereinstructed to complete the Listening TasksQuestionnaire (see Appendix 1) beforereturning for the next appointment Thiswas a 22-item questionnaire targetingloudness comfort and sound quality for avariety of stimuli and listening situationstypically encountered in the real world Sevenenvironmental sounds identified in thequestionnaire three loud sounds two averagesounds and two soft sounds were used toassess loudness comfort The Listening TasksQuestionnaire also included questionsregarding the subjectsrsquo own voice quality and

Table 1 Subject Hearing Aid and Earmold Information

Subject Current Yrs with Yrs of Use Hearing Aid Current Aid (Total) Earmold Vent Tubing

1 Diva 20 100 shell 24 mm 3 mm2 Claro 30 30 12 shell pressure 4 mm3 Diva 25 100 skeleton 15 mm 3 mm4 Canta 16 54 skeleton 40 mm 3 mm5 Diva 25 60 skeleton R-24 L-40 3 mm6 Diva 10 120 34 shell 40 mm 3 mm7 Diva 18 82 skeleton 40 mm 3 mm8 Diva 20 220 skeleton 24 mm 3 mm9 Diva 14 40 shell 24 mm 3 mm10 Claro 40 100 shell 18 mm 3 mm


use of the directional microphone and telecoilprograms The subjectsrsquo responses were usedas a guide for fine-tuning the frequencyresponse of the study hearing aid at thesubjectsrsquo fine-tuning appointments (see Figure2 for subject visit schedule) and to gathersubjective information regarding subjectsrsquoperceptions of their own voice quality TheListening Tasks Questionnaire was alsocompleted prior to returning for theevaluation appointments of each of the twoprocessing conditions to ensure subjects hadno loudness or sound quality concerns beforeobjective testing was performed

Subjects wore the hearing aids with theinitial fit for one week and returned for fine-tuning to address any subjective concernsidentified by the Listening TasksQuestionnaire Of the 10 subjects fourrequired fine-tuning one week following the32-channel signal processing fitting and sixrequired fine-tuning one week after the 64-channel signal processing fitting The mostcommon fine-tuning performed in each of theprocessing strategies was to reduce the high-frequency maximum gain at 6k Hz toeliminate feedback Following the one-weekfine-tuning appointment subjects were giventhe Listening Tasks Questionnaire andEnvironmental Sounds Questionnaire (seeAppendix 2) to complete before the next visit

The Environmental Sounds Questionnaireaddressed loudness comfort and satisfactionfor sounds such as car noise washingmachine phone ringing and so forth Thisquestionnaire differed from the ListeningTask Questionnaire in that it identified 18environmental sounds and subjects wereasked to assign an eight-point categoricalloudness scale and five-point satisfaction

rating for unaided and aided conditions Theeight-point loudness categorical scale was(1) did not hear (2) very soft (3) soft (4)comfortable but slightly soft (5) comfortable(6) comfortable but slightly loud (7) loud butokay and (8) uncomfortably loud The five-point satisfaction scale included (1) not goodat all (2) not too good (3) okay (4) pretty goodand (5) just right At this point subjects worethe aids for four weeks before returning formeasuring sentence recognition in quiet andnoise with the HINT and completion of theSubjective Loudness Test The protocol wasrepeated for the alternate fitting rationale

After subjects wore the hearing aids inboth processing strategies for five weeks thehearing aids were reprogrammed to the firstrandomly assigned rationale for another one-week trial The purpose of this crossoverdesign was to refamiliarize subjects with theprocessing strategy prescribed in the firsttrial period Following this one-week trialsubjects were asked to report any subjectivedifferences between the two processingstrategies

At the conclusion of the study thedirectional microphone performance of eachhearing aid was verified by measuring thefront-to-back ratio via probe-tube measuresin Program 2 To perform this measurementeach hearing aid was coupled to a nonventedearmold and placed into an artificial ear ina double-walled sound booth A 70 dB SPLANSI composite noise was presented at 0deg infront of the hearing aid and the output wassaved Then the hearing aid was rotated sothat the signal was at the maximum angle ofreduction and again the output was savedFigure 3 illustrates the mean front-to-backdifference and one standard deviation for


688

Figure 2 Subject visit schedule



689

the 20 study hearing aids across five discretefrequencies

HHeeaarriinngg IInn NNooiissee TTeesstt (HINT)

The HINT (Nilsson et al 1994) consistsof 250 sentences (25 lists of 10 sentences perlist) read by a male speaker The sentencesare of approximately equal length (six toeight syllables) and difficulty (first-gradereading level) and have been digitallyrecorded for standardized presentation TheHINT estimates the signal-to-noise ratio(SNR) at which the sentences embedded innoise can be repeated correctly 50 of thetime This type of measure is useful becauseit enables accurate reliable estimation ofspeech recognition in noise for context-richspeech materials

The administration of the HINT requirestwo lists to be presented (ten sentences each)for each experimental condition The firstsentence was presented 10 dBA below theattenuator setting necessary for the noise tobe presented at 65 dBA The first sentence isrepeated increasing the level of presentationby 4 dB until repeated correctly by the

subject Subsequently the intensity level isdecreased by 4 dB and the second sentenceis presented The stimulus level is raised(incorrect response) or lowered (correctresponse) by 4 dB after the subjectrsquos responseto the second third and fourth sentences Thestep size is reduced to 2 dB after the fourthsentence and a simple up-down steppingrule is continued for the remaining 15sentences The calculation of the SNRnecessary for 50 sentence recognition isbased on averaging the presentation level ofsentences 5 through 20 plus the calculatedintensity for the 21st presentation

HINT reception threshold for sentences(RTS) was obtained for two conditions foreach signal processing strategy (a) quiet and(b) diffuse R-Spacetrade restaurant noiseConditions were randomly assigned to avoidorder effects No subject received the samesentence list twice eliminating the potentialfor learning effects Before HINT testingbegan subjects were instructed to adjusttheir volume control to a comfortable level fora 65 dBA noise signal presented from aspeaker at 0deg at one meter

Figure 3 Mean front to back difference (in dB SPL) at 500 1000 2000 3000 and 4000 Error bars representplusmn1 SD


Recording the R-Spacetrade RestaurantNoise

A known noisy restaurant (noise floor of58 dBA at the recording position but thelevel of the noise created by the assemblageof people was significantly higher) withcarpeted floors wooden walls and a woodencathedral ceiling was secured for a privateparty The dimensions of the room where therecording was made were 36 feet (length) x36 feet (width) x 85 to 175 feet (height witha sloping roofline) Thus the volume of theroom was 22000 cubic feet The reverberationtime was unknown but is probably of limitedinterest here because the test materials(HINT sentences) were not spoken in therestaurant and therefore were not subject toany possible masking effects of reverberationFinally it was determined that the criticaldistance for the recording was about fivefeet Some of the tables (those nearest therecording position) were partially at or withinthe critical distance of the recordingmicrophones but many of the tables werebeyond Therefore the restaurant simulationwas a combination of direct and diffuseelements (L Revit personal communication)About 45 people were seated and servedbreakfast in the main seating area of therestaurant which when completely fullcould accommodate over 100 customers Atable at the center of the main seating areahad been removed and replaced by an arrayof recording microphones The eight mainrecording microphones were of the highlydirectional ldquoshotgunrdquo (interference-tube)variety typically used in the movie-makingindustry to record sounds from a distanceBecause each shotgun microphone had afrontal pick-up pattern spanningapproximately 45deg (plusmn225deg) around its axisthe eight microphones when placed in anequally spaced horizontal circular arraypicked up sounds arriving from all horizontaldirections around the center of the arrayThe presumed pick-up points (diaphragms)of the shotgun microphones were located twofeet from the center of the array A ninthomnidirectional microphone was placed atthe center of the array for calibrationpurposes

Each microphone was connected via apreamplifier to a separate track of amultitrack digital audio tape (DAT) recorder(Tascam DTRS system) In this way direct

and reverberated sounds were captured(recorded) from around the restaurant ldquoontheir wayrdquo to the center of the two-foot-diameter microphone array Later using theR-Spacetrade playback system in the laboratorythese ldquocapturedrdquo sounds were then releasedby the eight loudspeakers of the two-foot-diameter playback array In this way thesounds that had been captured at two feetfrom the center of the array in the restaurantwould now complete their paths toward thecentral listening position although now in adifferent time and place

Calibration of the R-Space RestaurantNoise

Before the recording of the breakfastparty calibration signals were recordedindividually through each microphone sothat playback levels could later be establishedto reflect the sound levels recorded in therestaurant Separately for each shotgunmicrophone an equalized loudspeaker (flatfrom 100 to 16000 Hz in 13-octave bands plusmn3dB) was held at a distance of two feet in frontof the diaphragm along the center of thepickup axis of the microphone A pink-noisesignal was delivered to the loudspeaker andadjusted to produce 84 dB SPL at the centerof the array For each shotgun microphonethe individual pink-noise calibration signalwas recorded onto the corresponding tapechannel In subsequent playback the gain ofthe amplifier for each R-Space loudspeakerwas adjusted to produce 84 dB SPL at thecenter of the loudspeaker array thusmirroring the calibration recording conditionOn average the sound pressure level of thebreakfast party as measured at thecalibration point in the restaurant was 75dBC or 72 dBA Therefore when properlycalibrated the playback system createdcorresponding average sound pressure levels

The HINT materials (sentences) and theldquoR-Space restaurant noiserdquo were transferredto a Macintosh hard drive using Toast 50software before being imported intoAudioDesk software Then in AudioDeskthe right track was separated from the lefttrack and the two tracks were digitallyspliced end-to-end to form one long ldquosoundbiterdquo This concatenated sound bite was thenrepeated as many times as was necessary toprovide noise long enough for the longestpresentation for the first HINT sentence


690



691

For subsequent HINT lists the same noisesound bite was used but with the startingtime differing from that of the previous listby several seconds Offset times of severalseconds were digitally edited and placed inthe appropriate channels thus producinguncorrelated noise Compton-Conley et al(2004 figure 4 p 447) recently reported thatthe long-term speech spectrum of the R-Spacerestaurant noise was very similar to the long-term speech spectrum of the HINT sentencesand noise

Figure 4 illustrates the signalpresentation system consisting of eightBoston Acoustics CR-65 loudspeakers(dimensions 257 mm x 162 mm x 200 mmfrequency response (plusmn3 dB) 65ndash20000 Hzcrossover frequency 4200 Hz woofer 135mm copolymer tweeter 20 mm domenominal impedance 8 ohms) placed in anequally spaced array at ear level one meterfrom the test subject in a 197 x 254 x 273meter double-walled sound suite (volume =1405 m3) with a reported reverberation timeof 019 seconds (personal communicationwith Industrial Acoustics Company) Theradius of the circle was one meter plus thedepth of the loudspeaker (200 mm)

Prior to testing two measurements weremade using narrow bands of pink noisecentered at 250 500 1000 2000 and 4000Hz from each of the eight loudspeakers Onemeasure was made at one meter and thesecond measure at a half a meter As expectedthe SPL measured at a half meter was 6 dB(plusmn1 dB) greater than the SPL measured at onemeter with the exception of 250 Hz for theloudspeakers at 45 90 270 and 317deg Thusfor the majority of loudspeakers andfrequencies between 500 and 4000 Hz thesubjects head was within the critical distancein this test environment Finally signals(sentences and noise) were fed from aMacintosh-driven digital audio workstationusing MOTO AudioDesk software and aMOTU Model 828 eight-channel FireWireAD-DA converter The 0deg loudspeaker wasdriven by an Alesis Model RA-150 amplifierin bridge-mono mode Individual channelsof Carvin DC-150 amplifiers drove theremaining loudspeakers

To ensure that the overall presentationlevel was 65 dBA for the noise condition a 5 in microphone connected to a Quest 1900precision sound level meter and OB-300 13-11 octave band (OB) filter was placed at ear

Figure 4 Illustration of the signal delivery and loudspeaker array used in the present study For the diffusecondition the noise was delivered from all eight loudspeakers and the HINT sentences were delivered fromthe loudspeaker at 0deg azimuth For the quiet condition the HINT sentences were delivered for the loudspeakerat 0deg azimuth


level with the subject absent one meter fromthe loudspeakers Because the noise fromeach loudspeaker was uncorrelated to eachother in the diffuse condition the outputlevel of each loudspeaker can be easilyadjusted to yield the same overall output foreach test-loudspeaker condition Calibrationof the loudspeakers was completed weeklyand the measured output was within plusmn1 dBof 65 dBA throughout the course of the studyFor the noise condition the overall outputfrom each loudspeaker was 56 dBA (10log10[8] where 8 denotes the number ofloudspeakers or 9 dB) Thus 65 dBA - 9 dBA= 56 dBA at each loudspeaker so whensummed the output from the eightloudspeakers at one meter was 65 dBA

The purpose for using this continuousnoise rather than the gated noise provided bythe HINT recording was that the noiseapproximates more closely many real-worldnoisy situations Finally a lavalieremicrophone was placed near the subjectrsquosmouth so the examiner could hear thesubjectrsquos response to the HINT sentences

SSuubbjjeeccttiivvee LLoouuddnneessss TTeesstt

The Subjective Loudness Test is madeup of 15 recorded environmental and speechstimuli derived from a Phonak sound compactdisc (CD) Included in the CD wererepresentative sounds of varying intensitiesand spectral properties Sound samplesincluded speech in quiet and backgroundnoise music and various environmentalsounds Fifteen stimuli of varying intensity (50 65 and 80 dBA) and spectralcharacteristics (low mid and high frequency)were chosen to be included in thequestionnaire (see Appendix 3) Subjects were seated in the middle of a 197 x 254 x273 meter double-walled sound suite with aloudspeaker at 0deg one meter away Aftereach sound sample was randomly presentedthe subject was instructed to assign a

loudness category and satisfaction categoryutilizing the Environmental SoundsQuestionnaire rating scale described earlierThe evaluation of loudness comfort andsatisfaction was completed only in theomnidirectional program

SSuubbjjeeccttiivvee LLoouuddnneessss TTeesstt Calibration

To assure that the presentation levelswere correct a 5 in microphone connected toa Quest 1900 precision sound level meterand OB-300 13-11 octave band filter wasplaced at ear level one meter from theloudspeaker A 50 65 and 80 dBA signal wasplayed through the audiometer via anexternal CD track The external output wasadjusted on the audiometer until the desiredsignal level was read on the sound levelmeter That position was then marked onthe external output control on the audiometerCalibration was completed weekly

RESULTS

Main Effect of Signal Processing

Figures 5 and 6 illustrate the mean RTS(in dB) for signal processing (32-channel and64-channel) and listening (R-Spacetraderestaurant noise quiet) conditions An RTSof 0 dB for the noise condition means thesubject required the intensity level of thesentences to be equal to the level of the noise(65 dBA) in order to correctly repeat back 50of the sentences Thus a higher RTS reflectspoorer performance and a lower RTS reflectsbetter performance For the quiet conditiona lower value represents the subject is ableto repeat back the sentence at a lowerintensity and therefore better performance

A repeated randomized block ANOVA(Kirk 1982) was performed on the dataappearing in Table 2 The ANOVA reveals nosignificant main effects for signal processing


692

Table 2 Mean Standard Deviation Standard Error Lower Boundary and Upper Boundary for HINTResults in the Noise and Quiet Conditions for 32- and 64-Channel Signal Processing

NOISE QUIET32 CHANNEL 64 CHANNEL 32 CHANNEL 64 CHANNEL (dB SNR) (dB SNR) (dB RTS) (dB RTS)

MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693

Figure 5 Mean reception threshold for sentences (RTS in dB) in quiet for 32- and 64-channel signal process-ing strategies in two listening conditions Error bars represent plusmn1 SD

Figure 6 Mean reception threshold for sentences (RTS in dB) in noise for 32- and 64-channel signal process-ing strategies in two listening conditions Error bars represent plusmn1 SD


in noise (F = 051 df = 19 p lt 083) or quiet(F = 288 df = 19 p lt 060)

Figure 6 illustrates that the meanperformance with 32-channel signalprocessing (36 dB) is not significantlydifferent than the mean performance for 64-channel signal processing (35 dB) whenlistening in diffuse noise Figure 5 illustratesthat the mean performance with 32-channelsignal processing (512 dBA) is also notsignificantly different than the meanperformance for 64-channel signal processing(504 dBA) when listening in quiet


Figure 7 illustrates the mean loudnessrating (plusmn1 SD) for the 32-channel and 64-channel signal processing strategies for the15 sound samples Figure 8 illustrates themean satisfaction rating (plusmn1 SD) for eachsignal processing strategy

Of the 15 sound samples 14 soundsreveal no significant difference in loudnessratings between channel conditions Onlyone sound sample demonstrates a significantdifference in mean loudness rating based on

a repeated randomized block ANOVA for 32-channel versus 64-channel signal processingAn ANOVA performed on the data appearingin Figure 7 illustrates the mean loudnessrating for the party noise (10) is significantlylouder for 32-channel signal processing (58)than the mean loudness rating for 64-channelsignal processing (47) (F = 58 df = 19 p lt04) Although a significant result is reportedthis finding needs to be viewed with cautiondue to the small effect size (11) Thecomputed observed power is 57 based upona computed alpha of 05 indicating that thesample size may not be sufficient for thereported size effect

An ANOVA performed on the dataappearing in Figure 8 illustrates 14 of the 15sound samples also show no significantdifference in satisfaction ratings Only themean satisfaction rating for the flute (5)sound sample for the 32-channel signalprocessing condition is significantly poorerthan the mean satisfaction rating for the 64-channel signal processing (42) (F = 76 df= 19 p lt 02) Again although a significantresult is reported this finding should beviewed with caution due to the small effect


694

Figure 7 Mean loudness ratings for 32- and 64-channel signal processing strategies for the Subjective Loud-ness Test Error bars represent plusmn1 SD



695

size (08) The computed observed power is 69based on a computed alpha of 05

EEnnvviirroonnmmeennttaall SSoouunnddss QQuueessttiioonnnnaaiirree

Figure 9 illustrates the mean loudnessrating (plusmn1 SD) for the 18 environmentalsounds for the unaided and the 32- and 64-channel aided conditions Because theEnvironmental Sounds Questionnaire wascompleted two times after wearing thehearing aids in each of the two processingconditions unaided data was collected twiceThe results of the two unaided conditionsare averaged in Figures 9 and 10 because nosignificant differences were found betweenthe two unaided conditions Figure 10illustrates the same results for thesatisfaction rating In Figures 9 and 10 if theANOVA performed on each environmentalsound between the unaided and the two aidedconditions is significant either a Dagger (p le 05)or DaggerDagger (p le 01) symbol is placed whereappropriate Further if a significantdifference is found between the two aidedconditions then an (p le 05) is placed whereappropriate

An ANOVA performed on each sound

sample revealed the mean loudness rating for17 of the 18 sound samples is significantlysofter in the unaided condition in comparisonto the aided 32-channel processing conditionOnly sound sample 9 chewing soft food (F= 46 df = 17 p lt 057) does not revealstatistical significance between aided andunaided conditions In addition the resultsof the ANOVA on the data appearing in Figure8 illustrate that the mean loudness rating forthe unaided condition is significantly softerthan the aided 64-channel condition for 17 ofthe 18 sound samples Only sound sample11 water boiling (F = 80 df = 16 p lt 104)does not show a significant difference betweenunaided and aided conditions

Results of the ANOVA also show theloudness rating on 17 of the 18 sound samples is not significantly different betweenaided conditions Only one sound sample 8(motorbike passing by) demonstratessignificant differences in loudness ratingsbetween the 32-channel and 64-channelconditions The mean loudness rating for the32-channel processing (62) is significantlylouder than the mean loudness rating forthe 64-channel processing (55) (F = 135 df= 19 p lt 025)

Figure 8 Mean satisfaction ratings for 32- and 64-channel signal processing strategies for the SubjectiveLoudness Test Error bars represent plusmn1 SD



696

Figure 9 Mean loudness ratings for the 18 environmental sounds unaided 32- and 64-channel signal processingstrategies for the Environmental Sounds Questionnaire Error bars represent plusmn1 SD

Figure 10 Mean satisfaction ratings for unaided 32- and 64-channel signal processing strategies for the Envi-ronmental Sounds Questionnaire Error bars represent plusmn1 SD


An ANOVA performed on each soundsample appearing in Figure 10 reveals themean satisfaction rating for the unaidedcondition is significantly poorer than theaided 32-channel processing aided conditionfor 6 sound samples 1 dog barking close by(F = 74 df = 13 p lt 025) 7 car turnsignal (F = 424 df = 13 p lt 002) 11water boiling (F = 39 df = 13 p lt 035) 15microwave oven beeping (F = 51 df = 13 p = 023) and 18 birds twittering (F = 54df = 13 p lt 007) Results of the ANOVA alsoreveal the mean satisfaction ratings for theunaided condition are significantly poorerthan the aided 64-channel processingcondition for 6 sound samples 7 car turnsignal (F = 424 df = 13 p lt 036) 11water boiling (F = 39 df = 13 p lt 016) 13telephone ringing (F = 39 df = 13 p lt 035)15 microwave oven beeping (F = 51 df =13 p lt 006) and 18 birds twittering (F = 54 df = 13 p lt 006)

Finally ANOVA results on the data inFigure 10 indicate the mean satisfactionrating for 17 of the 18 sound samples is notsignificantly different between aided channelconditions Only for sound sample 6 runningwater is the mean satisfaction rating for the32-channel processing condition (32)significantly poorer than the 64-channelprocessing condition (40) (F = 33 df = 13p lt 022)

The LLiisstteenniinngg TTaasskkss QQuueessttiioonnnnaaiirree

The Listening Tasks Questionnaire wasused to subjectively assess the subjectrsquosperception of his or her own voice quality Forthe 32-channel condition eight of the tensubjects report his or her voice to beldquocomfortablerdquo at one week postfitting Forthe 64-channel condition seven of the tentotal subjects report the sound of his or herown voice to be ldquocomfortablerdquo one weekfollowing the fitting Of the remainingsubjects the reports of own voice qualityincluded the descriptors ldquotinnyrdquo and ldquohollowrdquoYet increasing the number of channels doesnot seem to play a part in contributing to thenegative sound quality of the subjectrsquos ownvoice because those subjects who rated his orher voice to be other than ldquocomfortablerdquo in oneprocessing strategy also reported undesirabledescriptors of their voice in the alternatesignal processing rationale as well

DISCUSSION

Past research has questioned whether theadvantages of multichannel signal

processing can be realized without usersexperiencing the possible negative side effects(ie increased loudness discomfort decreasedspeech recognition and poor sound quality ofsubjectrsquos own voice) of increasing the numberof processing channels The current studyexamined the effect of increasing the numberof signal processing channels from 32 to 64channels on loudness comfort andsatisfaction sentence recognition and thesound quality of the subjectsrsquo own voice inADRO signal processing

As was previously discussed in theintroduction loudness discomfort as a resultof channel summation has been reported tobe a possible negative side effect whenincreasing the number of processing channelsThe results reported from the SubjectiveLoudness Test in the current study did notdemonstrate any significant differencesbetween the 32- and 64-channel processingin the loudness rating for 14 of the 15 soundsamples The mean loudness rating for onlyparty noise was significantly differentbetween channel conditions For this soundsample the party noise in the 32-channelcondition was reported on average to besignificantly louder than in the 64-channelcondition This finding was contrary to whatmight be expected if channel summation isoccurring However the overall meansatisfaction rating for the party noise soundsample was not significantly differentbetween channel conditions This indicatesthat even though the mean loudness of theparty noise might have been perceived to belouder with 32-channel signal processing incomparison to 64-channel signal processingsubjects overall were equally satisfied withthe loudness of the sound of the party noisein both conditions Therefore it seemed thatthe increase in loudness did not lead to adecrease in overall subject satisfaction orloudness discomfort for this sound sampleResults of the ANOVA on the EnvironmentalSounds Questionnaire also demonstratedthat the loudness ratings for 17 of the 18sounds were not significantly different Onlyone of the 18 sound samples (motorbikepassing by) showed any significant differencein loudness between the two channelconditions and again the satisfaction rating


697


between the 32-channel and 64-channelconditions was not significantly differentsuggesting no difference in loudnessdiscomfort Additionally this significantdifference in loudness and satisfaction ratingsmust be viewed with caution due to the smalleffect size between the two processingconditions as a result of the small number ofsubjects

It must also be noted that the lack ofreported differences in loudness ratings on theSubjective Loudness Test and theEnvironmental Sounds Questionnairebetween the two channel conditions may bea result of the fitting method used in thecurrent study Recall that the study hearingaids were fit using an in-situ loudnessjudgment method in which loudness comfortvalues were gathered across discretefrequencies for each channel condition Byutilizing the subjectrsquos individual dynamicrange as the target for the fitting theprobability of the subject experiencingloudness discomfort was decreased ascompared to using predicted values asdetermined by a prescriptive fitting targetTherefore any effects of channel summationmay have been accounted for at the time ofthe fitting

A second objective of this study was todetermine the effect of an increased numberof processing channels on speech recognitionA significant amount of research evaluatingthe effect of multichannel processing onspeech recognition has been performedutilizing different types of speech stimuli ina variety of testing conditions Results of theHINT in the current study demonstrate nosignificant differences in sentence recognitionin quiet or diffuse noise between the 32- and64-signal processing channel conditions Thiswas in agreement with past research usingHINT sentences measuring the effect ofmultichannel processing Moore et al (1999)reported no significant differences in HINTperformance when increasing the number ofsignal processing channels between one twofour and eight signal processing channelsYund and Buckles (1995) found when thenumber of signal processing channels wasgreater than eight channels increasing thenumber of processing channels had no effecton subjectsrsquo speech recognition abilities Eventhough the Moore et al (1999) and Yund andBuckles (1995) studies as well as the currentstudy utilized subjects with different levels

of hearing aid experience and different testingenvironments the results were stillcomparable This may be due to thesimilarities in subject inclusion criteria Allof these studies included users with mild-to-moderate hearing losses It is unknown ifsubjectsrsquo performance would be more affectedby increasing the number of processingchannels if subjects with more severe hearinglosses who may rely more significantly onthe temporal cues in speech were utilized inthe studies

Crain and Yund (1995) published resultsthat demonstrated that even as the numberof processing channels continues to increaseto thirty-one channels negative effects onspeech recognition were still not foundDespite the fact that Crain and Yund (1995)evaluated vowel and consonant stimuli andthe current study uses HINT sentences theresults were still in agreement that increasingthe number of processing channels did notcause speech recognition to decrease evenwhen the number of processing channelsexceeded 30 channels

Finally the current study also examinedthe effect of longer group delay on the soundquality of a subjectrsquos own voice as a result ofincreasing the number of processing channelsThe mean group delay for the 32-channeland 64-channel signal processing conditionswas reported to be approximately 7 and 13msec respectively in the current study Thesevalues did not exceed the 15 msec upper limitfor group delay for hearing-impaired listenersas previously reported by Stone and Moore(2005) Additionally the results of this studyindicate that increasing the number of ADROprocessing channels did not change thesubjectsrsquo subjective reports of own voicequality Those subjects who reported thesound quality of their own voice to beundesirable (ie tinny hollow etc) with oneprocessing strategy did with the otherprocessing strategy as well

These reports of a change in the soundquality of a subjectrsquos own voice also agreedwith data published by Stone and Moore(1999) who state that even delays shorterthan 10 msec could result in the subtle changeof the timbre of the subjectrsquos own voice as wasseen with some of the subjects in this studyAdditionally it was reported that subjectswho had a low-frequency pure-tone (PTAlf)average between 30ndash39 dB HL reportedsignificantly higher disturbance than the


698


other four groups with lesser or greaterdegrees of hearing loss In the current studythe mean PTAlf of the subjects was 386which would have included the subjects in thesignificantly more disturbed group in theStone and Moore (1999) study

Consideration must also be given to thefinding that increasing the number ofprocessing channels from 32- to 64-signalprocessing channels did not seem to yieldany significant advantages in the currentstudy as well As was reported previous data(Blamey et al 2004) has shown thatincreasing the number of processing channelscould be beneficial These advantages werenot realized in terms of improved loudnesscomfort and satisfaction improved sentencerecognition in quiet and noise or improvedsound quality of subjectsrsquo own voice in thisstudy Therefore it seems subjectsrsquoperformance with the 32-channel signalprocessing strategy was equivalent toperformance with the 64-channel signalprocessing in this study and the additionalchannels did not provide significantimprovement

It is worth noting that the results foundin this study are exclusive to the ADROtradesignal processing If another form ofprocessing (ie wide dynamic rangecompression linear with output limitingetc) would have been utilized the results ofincreasing the number of processing channelsis unknown and may not have been similarto what was published here Additionally ifan alternate fitting method were utilizedinstead of the manufacturerrsquos in-situ fittingprocedure the results might have varied aswell Additional research needs to becompleted to determine if alternative formsof signal processing or fitting methods wouldhave yielded different results between the 32-and 64-channel processing conditions It couldalso be beneficial if an experimental aid withfewer channels (ie one four or eightchannels) could be included in the futurestudies for a baseline comparison

In conclusion this study evaluated theeffects of an increased number of signalprocessing channels on loudness comfort andsatisfaction sentence recognition and soundquality of subjectrsquos own voice utilizing theADRO processing in 32-channel and 64-channel signal processing strategies in tensubjects The results of the study revealed

1 No significant differences in loudnesscomfort were present between 32-and 64-channel processing strategiesfor 14 of the 15 sound samples asmeasured by the Subjective LoudnessTest Additionally mean loudnessratings for 17 of the 18 sounds on theEnvironmental Sounds Questionnairerevealed no significant differencesbetween 32-channel and 64-channelconditions

2 No significant differences in loudnesssatisfaction were present between32- and 64-channel processingstrategies for 14 of the 15 soundsamples as measured by theSubjective Loudness Test Additionally mean satisfactionratings for 17 of the 18 sounds on theEnvironmental Sounds Questionnairerevealed no significant differencesbetween 32-channel and 64-channelconditions

3 No significant differences werepresent between 32- and 64-channelprocessing strategies in an adaptivedirectional microphone mode for thereception threshold for sentences(RTS in dB) on the Hearing In NoiseTest (HINT) sentences presented at0deg and diffuse R-Spacetrade noise (eightloudspeaker array) fixed at 65 dBA

4 No significant differences werepresent between 32- and 64-channelprocessing strategies in the adaptivedirectional microphone mode forreception threshold for sentences(RTS in dB) on the Hearing In NoiseTest (HINT) sentences presented at0deg in quiet

5 Subjective differences in soundquality of subjectrsquos own voice qualityare present between the 32- and 64-channel processing strategies asmeasured by the Listening TasksQuestionnaire

Acknowledgments The authors would like to thankDynamic Hearing Pty for providing a grant to coverthe direct costs involved for completing the projectand Interton for providing the hearing aids used inthe study Additionally the authors would especiallylike to thank Todd Fortune at Interton and PeterBlamey at Dynamic Hearing for their support andassistance throughout the project Finally the authorswould like to thank Michael Strube PhD Professorof Psychology at Washington University for his assis-tance in completing the statistical analysis and


699


appropriate interpretation of the data reported inthis manuscript

REFERENCES

Agnew J Thornton J (2000) Just noticeable and objec-tionable group delays in digital hearing aids J AmAcad Audiol 11330ndash336

American National Standards Institute (1996)American National Standard for Specification ofAudiometers (ANSI S36-1996) New York AcousticalSociety of America

Blamey P Martin LFA Fiket HJ (2004) A digital pro-cessing strategy to optimize hearing aid outputsdirectly J Am Acad Audiol 15716ndash728

Blamey P (2005) Adaptive dynamic range optimiza-tion a digital amplification strategy for hearing aidsand cochlear implants Trends Amp 9(2)T1ndashT22

Compton-Conley C Neuman A Killion M Levitt H(2004) Performance of directional microphones forhearing aids real world versus simulation J Am AcadAudiol 15440ndash455

Crain TR Yund EW (1995) The effect of multichan-nel compression on vowel and stop-consonantdiscrimination in normal-hearing and hearing-impaired subjects Ear Hear 16(5)529ndash543

Dillon H (2001) Hearing Aids New York ThiemeMedical Publishers

Frye G (2001) Testing digital and analog hearinginstruments processing time delays and phase meas-urements Hear Rev 10(8)36ndash42

Kiessling J Steffens T (1991) Clinical evaluation ofa programmable three-channel automatic gain con-trol amplification system Audiology 30(2)70ndash81

Kirk RE (1982) Experimental Design 2nd editionPacific Grove CA BrooksCole Publishing Company

Kuk F (2002) Considerations in modern multichan-nel nonlinear hearing aids Valente M ed HearingAids Standards Options and Limitations New YorkThieme Medical Publishers 178ndash213

Kuk F Ludvigsen C (2003) Changing with the timeschoice of stimuli for hearing aid verification pure-tones speech or composite signals Herersquos what touse and why Hear Rev 10(9)24ndash28 56ndash57

Martin LFA Blamey PJ James CJ Galvin KLMacfarlane D (2001) Adaptive dynamic range opti-misation for hearing aids Acoust Aust 29(1)21ndash24

Moore B Glasberg B (1986) A comparison of two-channel and single-channel compression hearing aidsAudiology 25(4)210ndash226

Moore B Peters RW Stone MA (1999) Benefits oflinear amplification and multichannel compressionfor speech comprehension in backgrounds with spec-tral and temporal dips J Acoust Soc Am105(1)400ndash411

Nilsson M Soli SD Sullivan J (1994) Developmentof the Hearing in Noise Test for the measurement of

speech reception thresholds in quiet and in noise JAcoust Soc Am 951085ndash1099

Stone MA Moore BCJ (1999) Tolerable hearing aiddelays I estimation of limits imposed by the audi-tory path alone using simulated hearing losses EarHear 20182ndash192

Stone MA Moore BCJ (2002) Tolerable hearing-aiddelays II estimation of limits imposed during speechproduction Ear Hear 23(4)325ndash238

Stone MA Moore BCJ (2005) Tolerable hearing-aiddelays IV effects on subjective disturbance duringspeech production by hearing-impaired subjects EarHear 26(2)225ndash234

Summerfield Q (1992) Lipreading and audiovisualspeech perception Philos Trans R Soc Lond B BiolSci 335(1273)71ndash78

Van Tassell D Solis D Kirby VM Widen GP (1987)Speech waveform envelope cues for consonant recog-nition J Acoust Soc Am 82(4)1152ndash1161

Yund EW Buckles KM (1995) Multichannel com-pression hearing aids effect of number of channelson speech in noise J Acoust Soc Am 97(2)1206ndash1223


700


Appendix 1 Listening Tasks Questionnaire

For the next week we would like you to listen to a number of different situations with your newhearing aids Your responses will help the audiologist to make any adjustments needed to help youhear better

Unless otherwise specified for all situations set the hearing aid on Program 1 and the volume ata comfortable listening level

Please check all that apply for each of the following questions

ENVIRONMENTAL SOUNDS

Loud sounds

bull When you hear a door slam the sound is loud but ok softer than expected slightly too loud other ____________________________________ uncomfortably loud ___________________________________________

bull When you hear a spoon drop into the sink the sound is loud but ok softer than expected slightly too loud other ____________________________________ uncomfortably loud ___________________________________________

bull When you hear traffic in the street the sound is loud but ok softer than expected slightly too loud other ____________________________________ uncomfortably loud ___________________________________________

Average sounds

bull When you hear the doorbell the sound is comfortable too soft too loud other ____________________________________ slightly too soft ___________________________________________

bull When you hear the phone ring the sound is comfortable too soft too loud other ____________________________________ slightly too soft ___________________________________________

Soft sounds

bull When you hear the sound of paper rustling the sound is comfortable too soft too loud other ____________________________________

___________________________________________

bull When you hear the sound of your own breathing it is comfortable too soft too loud other ____________________________________

___________________________________________


701


Feedbackwhistling

bull How often does your hearing aid whistle with the volume at your usual setting never constantly only when using the phone or wearing a hat other __________________________________ only when chewing or laughing ___________________________________________

bull Have you noticed any feedback with the volume full on never constantly only when using the phone or wearing a hat other __________________________________ only when chewing or laughing ___________________________________________

YOUR OWN VOICE

Listening to your own voice

bull When you are listening to your own voice it sounds comfortable hollow and echoing slightly too loud tinny slightly too soft other __________________________________

___________________________________________

PEOPLE TALKING

Listening to conversation in a quiet room

bull When you are talking to one other person speech sounds clear and comfortable too soft muffled or distorted too loud high pitched and tinny other __________________________________ hollow and dull ___________________________________________

bull When you are listening to the TV or radio speech sounds clear and comfortable too soft muffled or distorted too loud high pitched and tinny other __________________________________ hollow and dull ___________________________________________

NOISY PLACES

Listening to conversation in a noisy place

Set the hearing aid on Program 1 and the volume at a comfortable level and listen to the situationsbelow Then set the hearing aid on Program 2 and the volume at a comfortable level and listenagain in the situations below

bull When several people are talking in a group which program do you prefer to use I prefer to use Program 1 I prefer to use Program 2 Both programs work equally well


702


bull When someone is talking and there is background noise which program do you prefer to use I prefer to use Program 1 I prefer to use Program 2 Both programs work equally well

PROGRAMS

Changing programs

bull Do you have any problems changing the programs No I have tried using all programs Yes I find it difficult to change programs

bull Can you hear the beep when the program changes Yes I can hear a different number of beeps for each program No sometimes it is difficult to hear the beeps

VOLUME

Volume Control

bull How many times a day do you manipulate the volume control never 3ndash4 times 1ndash2 times 5 or more times

bull In what situations do you manipulate the volume control ______________________________ ______________________________ ______________________________ ______________________________

TELEPHONETelecoil

bull Have you tried using Program 3 the telecoil program when speaking on the telephone never 3ndash4 times 1ndash2 times 5 or more times

bull When you are listening on the telephone through Program 3 the speech sounds clear and comfortable too loud too soft other __________________________________

HOURS OF USE

bull On average how many hours per day did you use the experimental hearing aids in the last week never 7ndash8 hours 1ndash2 hours 9ndash10 hours 3ndash4 hours 11ndash12 hours 5ndash6 hours more than 12 hours

bull On average what percentage of the time did you use Program 2 in the hearing aid almost never 60 to 80 10 to 20 80 to 90 20 to 40 almost always 40 to 60


703


Appendix 2 Environmental Sounds Questionnaire

Name _______________________________ Date ________________

Office use only description of each hearing aidHA1 HA2

PART 1

During this week please listen to each of the following sounds with Program 1 of your hearing aidand without your hearing aid Please enter your responses to indicate the loudness of the sound andyour satisfaction with that loudness level for Program 1 of the hearing aid and without your hearingaid

For rating the loudness of the sound use the following loudness scale 7 = uncomfortably loud6 = loud but okay5 = comfortable but slightly loud4 = comfortable3 = comfortable but slightly soft2 = soft1 = very soft0 = do not hearx = donrsquot know eg did not encounter that sound

For rating your satisfaction with the loudness level use the following satisfaction scale5 = just right4 = pretty good3 = okay2 = not too good1 = not good at all

For example you might rate a particular sound as ldquovery softrdquo If ldquovery softrdquo is your preferred levelfor this sound then you would rate your loudness satisfaction as ldquojust rightrdquo If on the other handyou think the sound should be louder than ldquovery softrdquo then your loudness satisfaction rating mightbe ldquonot too goodrdquo or ldquonot good at allrdquo The loudness satisfaction rating is not related to how pleasingor easy it is to hear the sound but rather how satisfied you are with the loudness level perceived

IMPORTANT Remember to enter a loudness and satisfaction rating for Program 1 of the hearing aidsand without your hearing aids That means FOUR ratings for each sound described

Loudness scale 7 = uncomfortably loud6 = loud but okay5 = comfortable but slightly loud4 = comfortable3 = comfortable but slightly soft2 = soft1 = very soft0 = do not hearx = donrsquot know eg did not encounter that sound


704


Satisfaction scale5 = just right4 = pretty good3 = okay2 = not too good1 = not good at all

1 Dog barking close byWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

2 Traveling in a car with the windows closedWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

3 Traffic noise when standing on the curb of a busy roadWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

4 Your own breathingWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

5 Washing machineWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

6 Running water such as a toilet or showerWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

7 Car indicator signalWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

8 A motorbike passing byWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

9 Chewing soft foodWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

10 Vacuum cleanerWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

11 Water boiling on the stoveWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

12 Door slammingWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______


705


13 Telephone ringing close byWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

14 Refrigerator motorWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

15 Microwave oven beepingWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

16 Hair dryer or electric shaverWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

17 Lawn mowerWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

18 Birds twitteringWith hearing aid Loudness Rating _______ Satisfaction Rating _______Without hearing aid Loudness Rating _______ Satisfaction Rating _______

PART 2

Please answer the following questions by entering information on the line or ticking the relevant box

19 If you provided a low satisfaction rating (1 or 2) for some sounds in Part 1 of this questionnaire please provide reasons for your dissatisfaction

________________________________________________________________

20 What types of sounds or listening situations do you normally find loud or noisy________________________________________________________________

21 How often do you experience these loud or noisy sounds Several times per day Several times per week Only occasionally

A Question about the Multiprogram Hearing Aid22 Have you found sounds or listening situations that were too noisy or uncomfortably loud

Yes No

If yes which program would you prefer to use under these circumstances Program 1 Program 2 Program 3

23 For any of the programs do you have any other comments to make about how loud or soft soundsin the environment were or how you perceived loud and soft sounds


706


Appendix 3 Subjective Loudness Test

LOUD MEDIUM SOFT

LOW FREQUENCY SHIPrsquoS HORN DOUBLE BASS DISTANT THUNDERLOUDNESS (0ndash7) LOUDNESS (0ndash7) LOUDNESS (0ndash7)SATISFACTION (0ndash5) SATISFACTION (0ndash5) SATISFACTION (0ndash5)

MIDFREQUENCY TRAFFIC NOISE FLUTE FLOWING WATER LOUDNESS (0ndash7) LOUDNESS (0ndash7) LOUDNESS (0ndash7)SATISFACTION (0ndash5) SATISFACTION (0ndash5) SATISFACTION (0ndash5)

HIGH FREQUENCY WHISTLES KEYS BIRDS SING SOFTLOUDNESS (0ndash7) LOUDNESS (0ndash7) LOUDNESS (0ndash7)SATISFACTION (0ndash5) SATISFACTION (0ndash5) SATISFACTION (0ndash5)

BROADBAND PARTY NOISE DIALOG IN QUIET FEMALE SPEECH IN QUIETLOUDNESS (0ndash7) LOUDNESS (0ndash7) LOUDNESS (0ndash7)SATISFACTION (0ndash5) SATISFACTION (0ndash5) SATISFACTION (0ndash5)

ADDITIONAL LOUD PNEUMATIC HAMMER LOUD MUSIC PAPER RUSTLINGSOUNDS LOUDNESS (0ndash7) LOUDNESS (0ndash7) LOUDNESS (0ndash7)

SATISFACTION (0ndash5) SATISFACTION (0ndash5) SATISFACTION (0ndash5)


707

Effect of multichannel digital signal processing on loudness comfort sentence recognition and sound quality

Recommended Citation

tmp1316547776pdfeZP7s

Delivered by Ingenta to Washington University School of Medicine LibraryIP 1282521042 On Tue 20 Sep 2011 184911 681

J Am Acad Audiol 17681ndash707 (2006)

Department of Otolaryngology-Head and Neck Surgery Division of Adult Audiology Washington University School ofMedicine St Louis Missouri

Karen Mispagel Washington University School of Medicine Department of Otolaryngology-Head and Neck SurgeryDivision of Adult Audiology 660 South Euclid Ave Campus Box 8115 St Louis MO 63110 Phone 314-362-7490 Fax 314-747-5593 E-mail mispagelkentwustledu

Effect of Multichannel Digital SignalProcessing on Loudness Comfort SentenceRecognition and Sound Quality

Karen M Mispagel

Michael Valente

Abstract

This study evaluated the effect of increasing the number of processing channelsfrom 32- to 64-signal processing channels on subjectsrsquo loudness comfort andsatisfaction sentence recognition and sound quality of his or her own voiceTen experienced hearing aid users with mild-to-moderate sensorineural hearingloss wore behind-the-ear (BTE) hearing aids with Adaptive Dynamic RangeOptimization (ADROtrade) signal processing for a period of six weeks in the 32-channel and 64-channel conditions Results revealed no significant differencesin loudness comfort or satisfaction for the majority of sound samples asmeasured by the Subjective Loudness Test and Environmental SoundsQuestionnaire No significant differences in sentence recognition between thetwo processing conditions were found as measured by the Hearing In NoiseTest (HINT) Additionally no subjective differences in sound quality of subjectsrsquoown voice were determined by the Listening Tasks Questionnaire

Key Words Adaptive dynamic range optimization (ADRO) Hearing In NoiseTest (HINT) multichannel compression R-Spacetrade noise

Abbreviations ADROtrade = Adaptive Dynamic Range Optimization ANSI =American National Standards Institute BTE = behind the ear CLL = comfortablelistening level DSP = digital signal processing HINT = Hearing In Noise TestIRB = institutional review board MIL = most intelligible level OB = octave bandPTAlf = low-frequency pure-tone average RTS = reception threshold forsentences SNR = signal to noise ratio

Sumario

Este estudio evaluoacute el efecto de incrementar de 32 a 64 el nuacutemero de canalesde procesamiento de la sentildeal sobre el nivel agradable de intensidad subjetivadel sujeto y la satisfaccioacuten el reconocimiento de frases y la calidad del sonidode su propia voz Diez sujetos con experiencia en el uso de auxiliares auditivoscon hipoacusias sensorineurales leves a moderadas utilizaron auxiliaresretroauriculares (BTE) con Optimizacioacuten Adaptativa del Rango Dinaacutemico(ADROtrade) para procesamiento de la sentildeal por un periodo de seis semanasen condiciones de 32 y 64 canales Los resultados no revelaron diferenciassignificativas en el nivel confortable de intensidad subjetiva o en la satisfaccioacutenpara la mayoriacutea de la muestra de sonidos medidos por medio de la Pruebade Intensidad Subjetiva y el Cuestionario de Sonidos Ambientales Tampocode encontraron diferencias significativas en el reconocimiento de frases entrelas dos condiciones de procesamiento medidas con la Prueba de Audicioacutenen Ruido (HINT) Ademaacutes no se determinaron diferencias subjetivas en la



682











683











684










685











PROCEDURES

Subjects




686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






682











683











684










685











PROCEDURES

Subjects




686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






683











684










685











PROCEDURES

Subjects




686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






684










685











PROCEDURES

Subjects




686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






685











PROCEDURES

Subjects




686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






686









687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






687
















688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707












688




689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






689

















690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707













690



691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






691














RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707












RESULTS





692



MEAN 36 35 512 504SD 28 27 66 64SE 09 09 21 20LB 16 15 464 458UB 56 54 560 550



693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






693












694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707













694




695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






695










696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






696








DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707









DISCUSSION





697










698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707













698













699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707
















699



REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






REFERENCES

























700







Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707










Loud sounds




Average sounds



Soft sounds


___________________________________________


___________________________________________


701


Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707





Feedbackwhistling



YOUR OWN VOICE



___________________________________________

PEOPLE TALKING




NOISY PLACES





702



PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






PROGRAMS

Changing programs



VOLUME

Volume Control



TELEPHONETelecoil



HOURS OF USE




703



Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






Name _______________________________ Date ________________


PART 1








704
















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707



















705








PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707











PART 2



________________________________________________________________




Yes No




706



LOUD MEDIUM SOFT








707






LOUD MEDIUM SOFT








707




effect of multichannel digital signal processing on

Documents