Starkey Research & Clinical Blog

Some benefits of increasing the number of compression channels

Multichannel compression hearing aids: Effect of number of channels on speech discrimination in noise.

Yund, W.E. & Buckles, K.M. (1995).  Multichannel compression hearing aids: Effect of number of channels on speech discrimination in noise. Journal of the Acoustical Society of America 97(2), 1206-1223.

This editorial discusses the clinical implications of an independent research study. The original work was not associated with Starkey Laboratories and does not reflect the opinions of the authors. 

Yund and Buckles 1995 study of multichannel compression hearing aids offers valuable insight for the recommendation and fitting of current hearing aids.  A wide array of hearing aids are available, at different technology levels, and the number of compression channels is often a distinguishing feature. Typically, clinicians may recommend higher quality or more sophisticated processing for clients who lead active lives and participate in activities in challenging listening environments.  The effect of multiple compression channels on hearing aid performance in noise is therefore an important consideration in determining the most appropriate instrument for an individual client.

There are theoretical reasons to expect multiple channels of compression to improve or reduce speech recognition ability in the presence of noise.  Potential benefits of multichannel compression include plausible reduction of competing noise and preservation of audible speech information, as well as better fit to audiometric thresholds. Because the amplification within a given channel is based on the signal-to-noise ratio within that channel, it follows that speech information could be lost due to inadequate application of gain in instruments with fewer channels. Most previous studies of multichannel compression hearing aids compared their performance to that of linear hearing aids, but a study by Barfod (1978) showed improvement in performance as the number of channels increased from 2 to 4. At the time (1995) the reviewed paper was published, there were few studies to support performance increases with an increasing number of compression channels.

Yund and Buckles also acknowledged potential detrimental effects of multichannel compression, including reduction of temporal and spectral contrasts, resulting in increased phoneme confusion (Bustamante & Braida, 1987; Plomp, 1988, 1989; Villchur, 1989). The authors point out that theoretically, any deleterious effects of multichannel compression should increase with increasing number of channels, because systems with more channels would be better able to detect and respond to intensity variations across frequency. It follows, then, that at some point any benefit in multiple channel compression should plateau and possibly begin to decrease with increasing channels. Some have argued, however, that a reduction in amplitude contrasts might be counterbalanced somewhat by recruitment effects, which could result in larger perceptual contrasts (Moore, 1991).

Sixteen subjects with sensorineural hearing loss of varying degrees and etiologies participated in the experiment. The authors included participants with a broad range of hearing losses in an effort to examine any interactions between compression characteristics and hearing loss configuration or severity.  Test materials were based on the Nonsense Syllable Test (NST) by Resnick et al. (1975). The original test used nonsense syllables presented in a carrier phrase, but the present study used newly recorded stimuli presented in isolation. Test stimuli were CV or VC monosyllables that varied according to voicing (voiced or voiceless), consonant position (initial or final) and vowel context (/a/, /i/, or /u/).  All stimuli were recorded by a male speaker and a female speaker.

The nonsense syllables were combined with speech-shaped noise (French & Steinberg, 1947) prior to any linear or multichannel compression processing. The intensity of the noise was always 70dB, and speech was added to noise at 85, 80, 75, 70 and 65dB SPL to yield signal-to-noise ratios ranging from +15dB to -5dB.  Stimuli for each talker and each signal-to-noise ratio were presented in seven different processing conditions:  unprocessed, linear amplification (20 dB, flat), 4-channel compression, 6-channel compression, 8-channel compression, 12-channel compression and 16-channel compression. 

Not surprisingly, the results showed a strong effect of signal-to-noise ratio, with performance deteriorating as signal intensity decreased.  In quiet conditions, discrimination was better for the male voice than for the female voice, but this  reversed in the presence of noise, resulting in an advantage for discrimination of the female voice. The average spectrum for the female voice had relatively more energy in the high frequencies than the male voice and had peaks occurring at higher frequencies. The authors pointed out that because speech-spectrum noise has less energy in high frequencies, it may be less effective at masking the female voice, thus causing a reduced performance decrement for the female voice in the presence of noise.

There was a significant interaction between number of channels and voice; increasing the number of channels produced a greater improvement in discrimination for the male than it did for the female voice.  As noted above, speech-weighted noise has more energy in low frequencies, which could be more effective at masking the male voice. Because gain reductions in a multi-channel hearing aid are based on the level within a given channel, an instrument with fewer channels would be expected to lose more low-frequency speech information in its attempts to reduce gains applied to high-level, low-frequency noise. Instruments with a higher number of processing channels would therefore be expected to have less of a detrimental effect on low-frequency speech information, which is exactly what the results of the current study suggest.

There was a significant effect overall for the number of processing channels. Speech discrimination clearly improved as the number of channels increased from 4 to 8 channels and remained consistent above 8 channels.  There was no significant interaction between number of channels and signal-to-noise ratio; increases in the number of channels did not yield further improvement or decrement for performance at poor signal-to-noise ratios. 

The authors conducted a detailed analysis of consonant discrimination, which yielded information about the transmission of specific stimulus features and how they changed with increasing channels. In general, voiceless consonants were discriminated better than voiced, CV monosyllables were discriminated better than VC, and consonants in the context of /a/ were discriminated better than those in the context of /i/ or /u/. None of these features varied significantly as the number of channels increased. There was, however, a differential effect of number of channels on the perception of place and manner of articulation. Increasing the number of channels yielded more improvement for middle consonants than front or back consonants, and improved fricative perception than nasals and glides, and voiceless stops more than voiced stops. The most striking characteristic to benefit from increased channels was duration, which is a particularly important cue for differentiating fricatives.

The authors analyzed the frequency responses of the multi-channel instruments in their study and found that overall they were remarkably similar with one notable exception. Average amplification at 4 kHz and above increased as the number of processing channels increased from 4 to 8 to 16 channels. The improved high frequency response for instruments with more channels of processing, resulting in better  transmission of high-frequency speech cues is likely to help account for the noted improvements in consonant discrimination.

One aim of the current study was to determine any negative effects of increasing channels of compression. They found no negative effects, at least up to 16 channels, which was the maximum number used in the study. The authors mentioned a few contemporary studies that found increasingly negative effects with increasing number of channels for normal and hearing-impaired subjects, but only when high compression ratios were used (greater than 3:1).  It appears that for the compression characteristics used in the current study, any potential negative effect of increasing compression channels was negated by increases in the availability of speech information or possibly, as Moore suggested, recruitment effects.

Clinicians usually make recommendations for hearing aid style based on audiometric configuration, manual dexterity and anatomical variables, but the choice of technology level is often based on a patient’s lifestyle. There may be many reasons to recommend premium instruments for patients with active lifestyles, including more effective directional microphones, better automatic processing and more precise programming adjustments. The current study supports the importance of multiple channels of processing for better performance in noise, which is one of the major considerations for hearing aid users who participate in activities in challenging listening environments. Specifically, their study showed benefits for instruments with up to 8 channels of processing, a delineation that differentiates many entry-level hearing aids from their more sophisticated counterparts.

Yund and Buckles used multichannel compression instruments that were simpler than those available today.  Advances in digital processing have led to instruments that vary with regard to speed of processing, compression characteristics, adaptive directionality, noise reduction and other parameters.  Although an updated investigation of their hypotheses with current hearing aid technology could provide interesting insights into their findings, their study still supports the potential benefit of increased channels of processing for individuals with a wide range of hearing losses.

References

Barfod, J. (1978).  Multichannel compression hearing aids: Experiments and considerations on clinical applicability. In Sensorineural Hearing Impairment and Hearing Aids, ed. C. Ludvigsen and J. Barfod. Scandanavian Audiology Supplementum 6, 315-340.

Bustamante, D.K. & Braida, L.D. (1987). Principal-component amplitude compression for the hearing impaired. Journal of the Acoustical Society of America 82, 1227-1242.

French, N.R.  & Steinberg, J.C. ( 1947). Factors governing the intelligibility of speech sounds. Journal of the Acoustical Society of America 19, 90-119.

Moore, B.C.J. (1991). Characterization and simulation of impaired hearing: Implications for hearing aid design. Ear and Hearing 12, 154S-161S.

Plomp, R. (1988). The negative effect of amplitude compression in multichannel hearing aids in the light of the modulation-transfer function. Journal of the Acoustical Society of America 83, 2322-2327.

Plomp, R. (1989). “Reply to ‘Comments on ‘The negative effect of amplitude compression in multichannel hearing aids in the light of the modulation-transfer function.  [Journal of the Acoustical Society of America 83, 2322-2327.]’ [Journal of the Acoustical Society of America 86, 425-427].” Journal of the Acoustical Society of America 86, 428.

Resnick, S.B., Dubno, J.R., Hoffnung, S. & Levitt, H. (1975). Phoneme errors on a nonsense syllable test. Journal of the Acoustical Society of America 58, 114(A).

Villchur, E. (1989). “Comments on ‘The negative effect of amplitude compression in multichannel hearing aids in the light of the modulation-transfer function. ‘ [Journal of the Acoustical Society of America 83, 2322-2327.]” Journal of the Acoustical Society of America 86, 425-427.

Yund, W.E. & Buckles, K.M. (1995).  Multichannel compression hearing aids: Effect of number of channels on speech discrimination in noise. Journal of the Acoustical Society of America 97(2), 1206-1223.