Foundation for Hear. Aid Res., P. O. Box 306, Woodstock, NY 12498
Recent statements by the Food and Drug Administration imply that the only way a hearing aid can improve speech recognition in noise is to reduce the relative level of the noise. Hearing-aid circuits have not been designed that can effectively separate speech from noise, especially from the most common type of noise, competing speech. There is, however, experimental evidence---cited in this paper---that signal processing that restores lost speech cues to the hearing impaired can improve their speech recognition in any kind of noise, even though the signal-to-noise ratio is degraded. With such processing, hearing-impaired persons can better afford to lose cues to noise because they start out with more cues. They retain more of the redundant information in speech after masking has taken its toll, which helps them separate desired from undesired signals (a process that Broadbent calls, in normal listeners, ``selective listening.'') [D. E. Broadbent, Perception and Communication (Pergamon, New York, 1958)]. A recording of speech plus interference, processed to simulate recruitment, will be played both with and without compensation processing to illustrate how speech intelligibility can be improved in the face of increased relative noise.