As the only medical device that can restore hearing in deaf people, the cochlear implant has produced good speech recognition in quiet. However, the current implants are seriously limited in speech recognition in noise and in music perception. The long-term objectives of this research program are to understand signal processing in the normal auditory system and to restore functional hearing via auditory prostheses in hearing-impaired persons. Recent work from our laboratory and others has shown that pitch, temporal fine structure, and dynamic acoustic cues are crucial to improve realistic listening performance, but are not adequately encoded in current cochlear implants. Our working hypothesis is that extraction and encoding of these important cues will lead to an overall improvement in cochlear implant performance. We propose three novel methods to test this hypothesis. The three Specific Aims address each of these novel methods: (1) Co- vary stimulation rate and position to encode pitch; (2) Adapt modern vocoder algorithms to encode temporal fine structure; and (3) Use biologically-inspired signal processing to encode dynamic acoustic cues. Our multidisciplinary approach integrates psychophysical, speech coding, and signal processing techniques. A unique feature of this approach is that all algorithms are developed based on rigorous psychophysical and simulation measures, and will be evaluated and perfected in actual implant users with real-time implementations. Successful completion of the proposed research should yield results of high theoretical and practical significance. It will likely advance scientific knowledge on the centuries-old but still unresolved pitch coding question (Aim 1), bridge the technological gap between relatively rudimentary cochlear implants and modern telecommunication (Aim 2), and inspire translational work from basic research to clinical problems (Aim 3).
Showing the most recent 10 out of 34 publications
