The effect of multichannel compression on vowel and stop-consonant discrimination in normal-hearing and hearing-impaired subjects.

OBJECTIVE: Multichannel compression (MCC) processing can alter the speech spectrum, perhaps reducing spectral contrasts that are important for the discrimination of certain speech sounds. The effect of MCC processing on the discrimination of vowels and voiced stop consonants was studied. DESIGN: Vowels and voiced stop consonants were MCC-processed in two ways: 1) FLAT MCC having the same compression ratio in each channel, and 2)SHAPED MCC having compression ratios in each channel adjusted to the auditory area of the particular subject. The stimuli were processed both ways using 2, 4, 8, 16, and 31 independent compression channels. Unprocessed and linearly amplified stimuli were used as control conditions. Normal-hearing and hearing-impaired subjects were tested for changes in discrimination performance as a function of the MCC processing parameters. RESULTS: When the MCC processing was adjusted specifically for an individual hearing impared subject, no negative effect of increasing numbers of channels (2 to 31)was found. In the case of FLAT MCC processing, increasingly degraded discrimination performance was found for both subject groups as the compression ratio increased and as the number of channels increased. There was also a strong interaction between the effects of the number of channels and the compression ratio, with the negative effects of increasing numbers of channels being much greater at the highest compression ratio. CONCLUSIONS: Negative effects of MCC were found only for very extreme MCC conditions. MCC processing with compression ratios adjusted in each channel for the individual subject, and having as many as 31 channels, revealed no negative effects on vowel or voiced stop-consonant discrimination. These results do not support the prevalent view that MCC with more than two or three channels will be detrimental and should encourage further research on MCC processing with larger numbers of channels.

Effect of peak clipping on speech recognition threshold.

Speech recognition thresholds (SRTs) were measured for a closed set of spondees that had been processed by peak clipping and compression. Both hearing-impaired and normal-hearing subject groups showed progressively higher SRTs with increasing levels of peak clipping, with significant threshold shifts occurring for clipping levels greater than 18 to 24 dB. Neither subject group showed significantly elevated SRTs for the compression processed stimuli. Magnitude-squared coherence analysis of the speech stimuli revealed high levels of distortion generated by peak clipping and relatively low levels generated by the compression processing. Subsequent analysis suggested that the addition of distortion products and not the alteration of the speech waveform envelope was responsible for the observed threshold shifts, and that coherence analysis may be a valuable tool for predicting the effects of distortion on speech intelligibility. Judgments of sound quality showed that the clipping level where SRTs began to be significantly affected coincided with the clipping level at which the quality of the speech was judged to be unacceptable.

A Revised Speech Intelligibility Rating (RSIR) test: listeners with normal hearing.

Two experiments were conducted to examine the intelligibility of 72 passages of connected discourse prepared by Cox and McDaniel in their development of the Speech Intelligibility Rating (SIR) test. Intelligibility was assessed with a method-of-adjustment (MOA) procedure in which listeners adjusted the level of a multi-talker babble until they could just understand 50% of a passage; the measure of intelligibility was the signal-to-babble ratio, dB S/B. The objective was to develop a Revised Speech Intelligibility Rating (RSIR) test that would comprise a large number of equivalent passages that produce reliable intelligibility measures. In experiment 1, the S/B ratio was based on the overall root-mean-square (rms) levels of speech and babble, as represented by the average level of frequent peaks observed on a VU meter. Across all 72 passages, mean intelligibility was -1.43 dB S/B, and the measure of intelligibility for 42 passages was within +/- 0.5 dB of the overall mean for all 72 passages. In experiment 2, the S/B ratio was based on long-term rms levels of speech and babble measured in 16 one-third-octave bands, with center frequencies from 160 to 5000 Hz. In an effort to achieve greater equivalence in intelligibility among passages, the overall rms level of each passage was attenuated by the difference between SB16-band for an individual passage and S/B16-band for a reference passage. Mean intelligibility across all 72 passages was -8.06 dB, and the measure of intelligibility was within +/- 0.5 dB of the overall mean for 64 of the 72 passages. For those 64 passages, the 95% critical difference for five MOAs was 0.72 dB, which corresponds to an estimated percentage critical difference of 10.8%.

Noise reduction hearing aids: release from masking and release from distortion.

Automatic frequency response (AFR) hearing aids usually reduce their low-frequency gain in the presence of noise; several investigators have reported improved recognition of high-frequency speech information in low-frequency band-limited noise with AFR versus non-AFR hearing aids. In this work, masking patterns (masked threshold for frequency-modulated probe tones as a function of probe frequency) were obtained for a narrowband low-frequency noise. Speech recognition threshold for a set of high-frequency loaded monosyllables also was obtained in the presence of the same noise. Aided speech and masking pattern data for one normal and two hearing-impaired subjects wearing a master hearing aid incorporating a commercially available AFR circuit showed modest AFR effects. Moreover, masking noise spectra measured in ear canals of subjects wearing the master hearing aid showed evidence of substantial hearing aid-generated distortion products in the AFR-off condition. Results obtained from the normal subject listening with a low-distortion laboratory simulation of an AFR hearing aid showed greater release from masking for the same low-frequency attenuation as provided by the hearing aid. Improvements of speech recognition in noise observed with AFR hearing aids may result from some combination of release from upward spread of masking and reduction of distortion products generated by the hearing aid in the non-AFR setting.