Qualitative and quantitative results from a speech enhancement scheme assessed by hearing-impaired subjects

Abstract

A multi-microphone sub-band adaptive system for binaural preprocessing of speech signals for potential processing in future hearing aids has been investigated. Results are presented from both quantitative and qualitative assessments of the performance of the processing scheme as recorded from hearing-impaired subjects. The quantitative assessment is based on intelligibility testing of the algorithm using real-room recordings, T60=0.3 s, with multi-talker babble as the unwanted noise source. This evaluation has been conducted by using the four-alternative auditory feature test developed by Foster and Haggard [J. Br. Audiol. 21, 165–174 (1987)]. A corresponding mean opinion score on speech quality from each test subject provides the qualitative analysis. A response time measure was included with the qualitative measure. The processing scheme uses the least mean squares (LMS) [S. D. Sterns and R. A. David, Signal Processing Algorithms (Prentice–Hall, Englewood Cliffs, NJ, 1988)] adaptive noise cancellation algorithm in frequency-limited subbands. The inputs from each microphone are split into 16 contiguous subbands using a cochlear distribution according the function provided by Greenwood [J. Acoust. Soc. Am. 87, 2592–2605 (1990)]. Each frequency-limited subband is processed using a LMS adaptive noise cancellation filter operating in an intermittent or continuous mode depending on input signal characteristics. The results show a statistically significant and practically valuable improvement in both intelligibility and perceived speech quality using the proposed scheme.