Study Of Mandarin Chinese Perception In Cochlear Implant By Applying A Peripheral Auditory Model

Cochlear implant is the only effective medical device that can restore partial hearingto a profoundly or totally deafened person. It excites the residual auditory nerve fiberdirectly to make audition in accordance with the principle of functional electricalstimulation. With the development of bionic ear technology, cochlear implant has achievedhigh speech recognition rate. However, clinic reports show that there still exist manyproblems to be resolved. For most cochlear implant users, tone recognition rates are badand the music perceptions are worse. Moreover, in noisy environments, the auditoryperception performances of cochlear implant users decrease rapidly.This thesis first examined the conventional cochlear implant processing strategies andthe up-to-date research state. Then, focusing on improving Mandarin Chinese toneperception and quality of perception, the thesis did the study according to the auditorysystem model. Basing on the previous study, we analyzed the influence on mandarin toneperception depending on space coding and time coding respectively, and proposed a newcochlear implant speech processing strategy, which applied a peripheral auditory model togenerate stochastic nerve impulse trains based on time coding, and then studied somerelative problems. Finally, preliminary tests by using acoustic simulations were done toshow whether the processor with this strategy could get better Chinese speech perception.The thesis consists of following parts:(1) We tried to find out the influence on tone recognition when using space codingand time coding. The speeches were processed by CIS (Continuous Interleaved Sampling)strategy, ACE (Advanced Combination Encoding) strategy and recently proposedzerocrossing strategy which was one of non-uniform stimulation strategies, and speeches were analyzed after processed.(2) In order to present that speech information could beconveyed by impulse intervals, the stochastic impulse trains were generated by peripheralauditory model and the impulse trains would be analyzed.(3) In order to improve theChinese perception, this thesis applied the peripheral auditory model to generate stochasticnerve impulse stimulus to improve the modern speech processing strategy, trying totransmit more useful Chinese speech information to auditory center effectively andnaturally. And we tested the performance of proposed strategy under silence and noiseenvironment respectively.In summary, this thesis proposed a new non-uniform stimulation cochlear implantstrategy which was corresponding to normal mechanism of auditory system. Experimentresults presented that the Chinese speeches processed by proposed strategy could achievedbest quality and recognition rates. Finally, a future landscape of cochlear implant waspresented.