n\iDepartmentVeterans AffairsJournal of Rehabilitation Researchand Development Vol . 30 No . 1, 1993Pages 1—7Achieving prescribed gain/frequency responses with advances inhearing aid technologyCarol A . Sammeth, PhD ; David A . Preves, PhD ; Gene W . Bratt, PhD ; Barbara F . Peek, PhD;Fred H . Bess, PhDDepartment of Speech and Hearing Science, Arizona State University, Tempe, AZ 85287-0102;Argosy Electronics, Inc., Minneapolis, MN 55459; Audiology and Speech Pathology Service,Department of Veterans Affairs Medical Center, Nashville, TN 37203 ; Division of Hearing andSpeech Sciences, Vanderbilt University School of Medicine, Nashville, TN 37240Abstract—Technological limitations have restricted thecapability of older generation in-the-ear (ITE) hearingaids to closely match prescribed real ear gain/frequencyresponses . Newer technology, widely available in currently marketed ITE hearing aids, has considerablyimproved this capability . Data for 60 ears are presentedcomparing the real ear insertion gain (REIG) actuallyachieved to the target REIG, using ITE hearing aidshaving : 1) older generation narrow-band receivers, andamplifiers with single-pole-filter low frequency tone control and a class A amplifier output stage (n 30), and 2)newer generation amplifiers with a two- or four-pole-filterlow frequency tone control, and wide band receivers,containing a class D amplifier output stage (n 30).With the newer technology ITE hearing aids, the meansand ranges of deviation from target gain were reduced.Capability for achieving prescription REIG with ITEhearing aids can be further improved with multichannelamplifiers . Examples of the latter are shown for severaldifficult-to-fit audiograms.Key words : frequency responses, hearing aid technology,in-the-ear hearing aids, real ear gain.Address all correspondence and requests for reprints to : Carol A.Sammeth, PhD, Dept . of Speech and Hearing Science, Arizona StateUniversity, Tempe, AZ 85287-0102.Dr . Preves is associated with Argosy Electronics, Inc ., Drs . Bratt andPeek are with the DVA Medical Center, Nashville, and Dr . Bess is withVanderbilt University School of Medicine.This work was supported in part by the Department of Veterans Affairs,Rehabilitation Research and Development Service .INTRODUCTIONOver the past 15 years, numerous prescriptionformulas for hearing aid fitting have been proposed,each specifying a desired gain/frequency responsebased on the audiologic test results of a givenhearing-impaired patient (1,2) . The majority ofthese formulas use audiometric thresholds and/orsuprathreshold loudness judgments in their calculations, with the most common goal that of maximizing audibility of the speech spectrum or of amplifying the speech spectrum to "most comfortable"listening level . Given the current wide availabilityand use of probe microphone measurement systems,prescriptive fitting of the real ear insertion response(REIR) appears to have largely replaced the traditional, comparative speech approach (3), at least forinitial selection of amplification parameters.Unfortunately, a problem has arisen in thepractical application of prescriptive approaches tohearing aid fitting . In-the-ear (ITE) hearing aids,which make up the largest percentage of hearingaids sold today (4), have been reported to be tooinflexible to adequately match prescribed gain/frequency responses (5,6) . The primary reason forthese negative findings to date has been limitationsin past hearing aid technology . Recent advances intechnology, however, may make closer approximation to prescribed gain/frequency responses possible .
2Journal of Rehabilitation Research and Development Vol . 30 No . 1 1993Past Limitations and New Advances in TechnologyTwo major limitations in hearing aid technology have been additive in causing the limitedcapability of many older technology single-channelITE hearing aids to provide a good match toprescribed frequency response:1.2.Older-vintage hearing aid receivers had a primary frequency response peak in the 1600 to2200 Hz frequency range and inadequate highfrequency sensitivity . The use of these receiversfrequently resulted in two problems . First, alarge dip was often seen in the REIR in thefrequency range of the peak in the real earunaided frequency response (REUR), usually atabout 2800 Hz . Second, there was inadequatehigh frequency REIR above about 3000 Hz.Tone controls available had a limited range ofvariability . Until recently, most ITE hearingaids had low frequency tone controls comprisedof a single-pole highpass filter, primarily because of physical space limitations preventingthe packaging of additional capacitors requiredfor more filter poles . These low frequency tonecontrols were quite limited in their ability tovary the frequency response, having a slope ofonly 6 dB per octave . For example, it waspossible to reduce gain at 500 Hz by only about10 dB without changing the high frequencygain significantly.Because of these limitations, a variety of methods were used in the attempt to fine-tune thefrequency response of these ITE hearing aids tobetter match the prescription target . One techniqueused was to overfit the overall gain of the hearingaid in order to achieve the required high frequencygain and then to rely on varying the low frequencytone control to match the prescribed gain values atlow and mid frequencies . However, since the lowfrequency tone controls had such a limited range,the resulting frequency response often still exceededtarget gain in the low and mid frequencies . Thedesired high frequency insertion gain target wasseldom achieved anyway, even with the highestpossible overall gain provided . This was especiallytrue for patients with steeply sloping audiogramsand for those with hearing loss confined to the highfrequencies.Limited improvement was sometimes obtainedwith electronic shifting of the peak in the frequencyresponse of the receiver to a higher frequency . Thiswas accomplished by placing an appropriate-valuecapacitor across the receiver and/or by using aconstant voltage rather than a constant-currentamplifier output stage . Some hearing aid designersalso turned to mechanical/acoustical techniques toincrease high frequency gain relative to low and midfrequency gain . These included using a low-cutresponse microphone and a stepped bore earmold inconjunction with damping in the hearing aid receivertubing (7) . The stepped or flared earmold boreproduces a miniature megaphone in order to matchthe high acoustic source impedance of the receiver tothe low acoustic impedance of the ear canal at thehigh frequencies . When combined with the newer,wider-band receivers, this approach also achievedsome improvement in frequency response fitting.More recent ITE hearing aids that are on themarket employ "active" tone controls, consisting oftwo- or even four-pole filters, having slopes of 12dB/octave and 24 dB/octave, respectively . Thesehigher-order filters provide much greater flexibilityin varying the frequency response . It is now notuncommon to have low frequency tone controls witha 40 dB range at 500 Hz.In the past, most ITE hearing aids utilized aclass A power output stage in the amplifier . Although not directly related to the number of filterpoles, hearing aids with a class A output stage areoften associated with a narrow-band receiver response and limited tone control flexibility . ManyITE hearing aids currently being produced employeither a class D or class B output stage in theamplifier and a receiver with a wide-band frequencyresponse . These newer hearing aids typically have aprimary frequency response peak at about 3000 Hzand provide adequate gain in the high frequencies.Thus, many ITE hearing aids with good insertionfrequency responses are associated with amplifiersthat have class D and class B output stages.METHODComparison of Older Versus Newer TechnologyHearing AidsIn order to illustrate the degree of improvementseen in frequency shaping flexibility with olderversus newer single-channel technology, data arepresented here for 30 ears fit with older-technology
3Section 1 . Digital Techniques in Acoustic Amplification : Sammeth et al.ITE hearing aids that used linear, class A amplifiersand relatively narrowband receivers, and 30 ears fitwith newer-technology class D amplifiers with widerband receivers . The hearing aids with class Aamplifiers (standard linear, from Argosy Electronics, Inc ., Minneapolis, MN) used single-pole filtersfor the low-frequency tone control . Five of the classD amplifiers had low-frequency tone controls comprised of two-pole filters (Argosy Linear Plus ) andthe remaining 25 were comprised of four-pole filters(Argosy Manhattan li ). Due to the small samplesize for the two-pole class D amplifier, the data arepresented as a group . Examination of individualdata, however, revealed no apparent difference inresults across the two class D devices.These data were obtained retrospectively frompatient files of veterans who received hearing aids atthe Department of Veterans Affairs Medical Centerin Nashville, TN, and represent, therefore, typicalclinical accuracy in routine hearing aid fittings . Thedata shown for the older technology ITEs werecollected in late 1989, and the data shown for thenewer technology ITEs were collected in late 1991 bythe same two audiologists . Shown in Table 1 are themean audiograms, and standard deviations, for theears fit with newer versus older technology hearingaids . Audiometric configurations within each groupranged from mildly to severely sloping.The following protocol was used for all fittings.For each ear, desired gain/frequency response wascalculated using the NAL-R—revised NationalAcoustics Laboratories formula (8) and a customITE was ordered . t With an order, the manufacturer'It is notable that current prescriptive formula approaches were notspecifically developed for use with adaptive-frequency-response (AFR)hearing aids such as the Argosy Manhattan II or K-Amp (EtymoticResearch, Inc ., Elk Grove Village, IL) . In fact, the prescribed frequencyresponse is an attempt to provide the best compromise when a single,fixed frequency response must be used across multiple listeningenvironments . Unfortunately, however, there is no established approachfor the setting of electroacoustic parameters in AFR hearing aids . Forour purposes, we chose to set the gain/frequency response of the AFRhearing aids to match prescribed values with a moderate speech-levelinput (70 dB SPL), on the assumption that adaptive processing willoccur at higher and lower input levels.An argument can also be made for specifying different frequencyresponses for AFR hearing aids than those prescribed with currentformulas . Take, for example, an AFR hearing aid that functions in amanner similar to the Argosy Manhattan II ; that is, low frequency gainis reduced as input level increases . The assumptions underlying thisapproach are that high-level, low frequency energy is more likely to benoise than speech, and that if excessive upward spread of maskingoccurs in a hearing-impaired ear, it will be reduced . It can be arguedthat more gain should be supplied in the low frequencies for this type ofprocessor than is currently prescribed, particularly when fitting patientsTable 1.Means and standard deviations of audiologicalthresholds for each group of ears.250500Frequency (Hz)1000 2000 33(14)39(14)53(13)63(19)71(19)73(21)x meansd standard deviationwas given only the earmold impression and theprescribed full-on 2 cc coupler gain values.Audiograms were not supplied . To provide maximum flexibility in fitting to prescribed values, eachhearing aid was ordered with a tone control andvariable venting inserts . In addition, most of theManhattan II hearing aids had a bandpass trimpot.When a hearing aid arrived from the manufacturer,electroacoustic evaluation was accomplished first toensure proper functioning (ANSI S3 .22 - 1987), thenthe hearing aid was fit to the patient's ear usingprobe-microphone measurement of real ear insertiongain (REIG) with a Fonix model 6500 instrument . Abroadband composite signal was presented at 70 dBSPL from a loudspeaker positioned 1 meter fromthe subject at a 45 azimuth . Trimpots and/orventing were adjusted as necessary to achieve theclosest possible match to NAL-R prescribed real earinsertion gain (REIG) values at each frequency . Forthe data presented here, the REIG values obtainedwere compared to the insertion gain values that hadbeen prescribed, in order to determine the degree ofdeviation from target.RESULTSThe means of the 30 prescribed REIGs at eachfrequency (filled circles), and the means of the 30with normal or near-normal thresholds in the lower frequencies . Therationale is that low frequency gain will be sufficiently reduced inhigh-level noise environments by the adaptive function to provide theabove benefits, but that the greater low frequency gain supplied inlow-level, quiet environments will actually enhance perceptual soundquality (10) . It is clear that further research into the development ofappropriate fitting techniques for AFR hearing aids is needed . Atpresent, however, we would encourage the use of broadband stimulipresented at multiple input levels to more fully characterize thefunctioning of these devices (11) . In addition, evaluation of speechrecognition in quiet and in noise is important, and adequate follow-up isa crucial factor in assuring user success.
4Journal of Rehabilitation Research and Development Vol . 30 No . 1 1993REIGs actually obtained for each fitting (opencircles), are shown in Figure la for the older-vintagehearing aids, and in Figure 2a for the newertechnology hearing aids . Data at 250 Hz wereeliminated from the figures due to noise problems inthe real ear measurements at this frequency . Shownin Figure lb and Figure 2b, for the older and newertechnology ITEs, respectively, are the means andstandard deviations of individual deviations fromtarget ; that is, the differences between prescribedinsertion gain and the insertion gain actually obtained with each hearing aid . A positive valueindicates that the gain obtained was greater thanthat prescribed and a negative value indicates thatthe gain obtained was less than that prescribed.Linear regression lines and values are also shown inFigure lb and Figure 2b.Note that the older technology hearing aids(Figure 1) tended to provide too much gain throughthe mid frequencies and too little gain in the highfrequencies relative to prescribed values, as reflectedboth in the mean data and in the downward slope ofthe regression line in Figure lb . In fact, adequategain was achieved at 4000 Hz in only two of the 30ears, and too much gain resulted at 1500 Hz in 24 ofthe fittings.In contrast, the newer technology hearing aids(Figure 2) provided, on average, closer approximations to prescribed REIG across the frequencyrange . In particular, the flatter regression line, andbetter fit to the mean data, indicates an improvement in the ability to achieve sufficient gain at 4000Hz without excessive mid-frequency gain . Smallerstandard deviations indicate the reduced spread inthe data for newer technology hearing aids.Another way to measure the central tendency ofthe data is to examine the number of fittings thatfall within an acceptable degree of error from targetREIG . We generally consider plus or minus 5 dBfrom target to be acceptable . Table 2 lists, for olderversus newer technology hearing aids, the percentageof fittings at each frequency in which the REIGactually fit was within 5 dB of the prescribedREIG. Across mid and high frequencies, a substantially higher percentage of fittings fell within theseguidelines for the newer technology ITEs than forthe older technology ITEs, again indicating a largeimprovement in frequency response shaping capability .a30M25C 20w15z 105050010004000FREQUENCY (in Hz)Figure la.Closed circles represent the mean prescribed insertion gain/frequency response, and open circles represent the mean REIGactually achieved, for the 30 older technology ITEs.a30— 25CO15550010004000FREQUENCY (in Hz)Figure 2a.Closed circles represent the mean prescribed insertion gain/frequency response and open circles represent the mean REIGactually achieved, for the 30 newer technology ITEs.Multi-Channel AmplifiersHistorically, most hearing aid amplifiers havebeen single-channel devices . In order to compensatefor the insertion gain dip at 3000 Hz, some
5Section I . Digital Techniques in Acoustic Amplification : Sammeth et al.Table 2.Percentage of ears in each group with deviation fromtarget REIG of plus or minus 5 dB or less.b5aFrequency (Hz)1000 1500 ENCY (in Hz)Figure lb.The means and standard deviations of individual deviationsfrom target for each of the 30 older technology ITEs (i .e ., thedifferences between prescribed insertion gain and the insertiongain actually obtained) . A positive value indicates that the gainobtained was greater than that prescribed and a negative valueindicates that the gain obtained was less than that prescribed.The linear regression line is also shown.201050-10-15-2050010004000FREQUENCY (in Hz)Figure 2b.The means and standard deviations of individual deviationsfrom target for each of the 30 newer technology ITEs . Apositive value indicates that the gain obtained was greater thanthat prescribed and a negative value indicates that the gainobtained was less than that prescribed . The linear regression lineis also shown .manufacturers have successfully employed a secondchannel of amplification via an additional bandpassfilter to add gain in the 3000 Hz frequency range.An example of this approach is shown in Figure 3a.The effect of the second channel on the HA-1 2 cccoupler frequency response is shown in Figure 3b.The effect on frequency response measurements in areal ear, obtained with an Acoustimed HA-2000, isshown in Figure 3c . A significant increase in midand high frequency REIG is seen with counterclockwise rotation of the bandpass potentiometer in thisITE hearing aid.One of the most recent technological advancesin hearing aids has been incorporation of a graphicor parametric equalizer within the hearing aidamplifier . Multichannel amplifiers give much finerresolution in frequency response shaping than dosingle-channel amplifiers . With a graphic equalizer,gain is controllable in each frequency band . With aparametric equalizer, gain is controllable in eachfrequency band, but, in addition, the crossoverfrequencies of the bands can be shifted.The ability to precisely match a prescribedgain/frequency response with multichannel devicesgreatly exceeds that of older single-channel devices.Three individual examples of matches between measured and NAL-R prescribed REIG with a 3-channelparametric equalizer ITE hearing aid (Argosy 3Channel-Clock ) are shown in Figure 4 . Figure 4aillustrates a match obtained for a severely slopingaudiogram with normal hearing through 2000 Hz.This audiometric configuration is particularly difficult to fit with single-channel hearing aids becausethe actual insertion gain achieved will often be lessthan the target gain prescribed in the high frequencies, and greater than that prescribed in the midfrequencies . Figure 4b illustrates the match obtainedfor an audiogram with a rising configuration.
6Journal of Rehabilitation Research and Development Vol . 30 No . 1 1993matched the prescribed REIG for all three cases withthe 3-channel parametric equalizer ITE hearing aid.DISCUSSION4030-20b.a-10 l- --100-)ME -205002002 505005k 8k1k2kFrequency (Hz)1k2ki4k8HMIMI504030 ro-'o20 010 Q.MIMMI-10Frequency (Hz)Figure 3.Figure 3a illustrates an example (Argosy Manhattan II) of theuse of an additional amplifier channel with bandpass filter,added to a single-channel amplifier to obtain an increase in gainin the mid to high frequency region . The effect on 2 cc couplergain of adding the second channel is shown in Figure 3b, andwith probe microphone measurement of REIG in Figure 3c.Finally, Figure 4c illustrates a "reverse cookie-bite"configuration ; that is, a region of normal hearing inthe mid frequencies with hearing loss confined tohigher and lower frequencies . For this latteraudiogram, it is typically impossible to achieve areasonable approximation to prescribed gain usingsingle-channel ITE technology, and possible, butdifficult, with acoustic earmold or earhook modifications in behind-the-ear hearing aids . Note theaccuracy with which the gain/frequency responseThe data shown indicate substantial improvement in the capability of newer technology ITEhearing aids to achieve good approximations toprescribed gain/frequency responses . With two- orfour-pole filters for low frequency tone controls,and wide-band receivers with class D amplifieroutput stages, adequate high frequency gain can beachieved without overamplification of the mid frequencies . Although the hearing aids used in thisstudy were from Argosy Electronics, these technological innovations are also available from othermanufacturers and we would expect to see comparable results . The flexibility of the three-band ITEhearing aid with parametric equalization is sufficientto supply an accurate match even with more unusualaudiograms that have previously been quite difficultto fit.This improved flexibility also provides greateropportunity for successful revision of a patient'sgain/frequency response, if this is considered desirable after the initial fitting to prescribed values.Because prescriptive formulas are based on meandata for several parameters, the optimal frequencyresponse for an individual patient may in fact differfrom the prescribed target . A number of researchershave argued that a prescribed gain/frequency response should be considered only as a "ballpark"starting point in the fitting process, followed byadjustment based on evaluation of speech understanding ability and/or perceptual sound quality,and with follow-up regarding patient satisfaction(8,9) . Modifications to the prescribed frequencyresponse for a patient whose speech recognitionperformance is poorer than expected would typicallybe in the direction of increased high frequency gainrelative to low frequency gain . The data presentedhere suggest that such modification will be achievedmore easily with newer technology than it has beenin the past.ACKNOWLEDGMENTSThe authors would like to thank Todd L . Fortune,PhD, for assistance with collection of some of the datapresented in this manuscript.
7Section I . Digital Techniques in Acoustic Amplification : Sammeth et al.c.a.Frequency (Hz)i!iiIIIi2501020302505001000200040008000 40125250Frequency (Hz)100020005004000800000200 40CO 5060m INI1111111MIMIMMIll-90 100Frequency (Hz)125 250 500 1000 2000100110110250Frequency (Hz)4k5001k2k8k-- 11M5040250mFrequency (Hz)500 1k2k4k8k50. . . 40a30g10 %).2010%3020-M61 all 0-10wFrequency (Hz)250 5001k2k.- 1MIllill 0 w-. - -10Figure 4.Three examples of the match to prescribed REIG target obtained with a 3-channel ITE hearing aid with parametric equalization(Argosy 3-Channel-Clock) . For each case, the audiogram is shown in the upper part of the figure, and REIG measurements in thelower part of the figure, with the smoother line representing the target insertion gain .REFERENCES1 . McCandless GA . Hearing aid formulae and their application . In : Sandlin RA, editor . Handbook of hearing aidapplication (vol. I) . Boston : College-Hill Press, 1988:221-38.2 . Humes L . Prescribing gain characteristics of linear hearing aids . In : Studebaker G, Bess F, Beck L, editors . TheVanderbilt hearing aid report II . Parkton, MD : YorkPress, 1991 : 13-21 .3 . Carhart R . Tests for the selection of hearing aids.Laryngoscope 1946 ;56 :780-94.4 . Cranmer K . Hearing instrument dispensing - 1991 . HearInstr 1991 ;42(6) :6-13 .5 . Mueller HG, Bryant M, Brown W, Budinger A . Hearingaid selection for high-frequency hearing loss . In:Studebaker G, Bess F, Beck L, editors . The Vanderbilthearing aid report II . Parkton, MD : York Press, 1991 :269-86 .6. Sammeth C, Peek B, Bratt G, Bess F, Amberg S . Abilityto achieve gain/frequency response and SSPL-90 underthree prescription formulas with custom in-the-ear hearingaids . J Am Acad Audiol . In press.7. Knowles H, Killion M . Frequency characteristics of recentbroad-band receivers . J Audiolog Technique 1978;17(3) :86-9.8. Byrne D, Dillon H . The National Acoustic Laboratories(NAL) new procedure for selecting the gain and frequencyresponse of a hearing aid . Ear Hear 1986 ;7(4) :257-65.9. Bratt G, Sammeth C . Clinical implications of prescriptiveformulas for hearing aid fitting . In : Studebaker G, BessF, Beck L, editors . The Vanderbilt hearing aid report II.Parkton, MD : York Press, 1991 : 23-34.10. Punch J, Beck E . Low-frequency response of hearing aidsand judgments of aided speech quality . J Speech HearDisord 1980 ;45 :325-35.11. Preves D, Beck L, Burnett E, Teder H . Input stimuli forobtaining frequency responses of automatic gain controlhearing aids . J Speech Hear Res 1989 ;32 :189-94.
University, Tempe, AZ 85287-0102. Dr. Preves is associated with Argosy Electronics, Inc., Drs. Bratt and Peek are with the DVA Medical Center, Nashville, and Dr . Bess is with Vanderbilt University School of Medicine. This work was supported in part by the Department of Veterans Affairs, Reha