Enhancing Tone Recognition for Cochlear Implant Users

点击翻译按钮获取中文摘要

The overall long-term goal of this research is to develop novel speech processing strategies for cochlear implant users speaking tonal languages. In attempt to improve tonal language perception in cochlear implants, the present proposal focuses on enhancing the transmission of tonal information in electric hearing, because tonal information is crucial for speech perception in tonal languages, and that current cochlear implants do not provide sufficient tonal information for users. Three experiments are proposed to enhance different tone- related information in cochlear implants. Specific Aim 1 is to investigate whether pre-processors modifying single syllable's overall amplitude envelope to better resemble its fundamental frequency contour can enhance tone recognition in cochlear implants. Pilot studies revealed that such amplitude envelope modifications significantly improved Chinese tone recognition in 4-channel cochlear implant acoustic simulations, without adversely affecting vowel recognition. Tone and vowel recognition of tonal-language-speaking cochlear implant users will be tested under different amounts of such amplitude envelope modifications, to search for the best trade-off between tone recognition improvement and possible vowel recognition decrease. Specific Aim 2 is to investigate tonal language perception in speech-shaped noise or speech-babble noise with cochlear implant recipients using a hearing aid in the non-implanted ear. Pilot studies using acoustic simulations of a cochlear implant with a hearing aid showed that, Chinese sentence recognition in speech- shaped noise was significantly enhanced as more low frequency information was added; specifically, such sentence recognition enhancement was mostly due to better tone recognition when the low frequency information was 500 Hz. The effects of frequency range and signal level of the low frequency acoustic hearing restored by hearing aid, as a complement to the contra-lateral electric hearing, will be investigated. Specific Aim 3 is to investigate whether tonal information can be effectively encoded using extra low pitch sensations produced by wide bipolar stimulation modes in cochlear implants. Pilot studies using 4-channel cochlear implant acoustic simulations revealed that Chinese tone recognition was significantly enhanced as the number of extra frequency channels in the pitch range was increased. Device implementations and applications of mixed stimulation modes for tonal language speech processing in cochlear implants will be investigated.