Signal Processing Strategies on Cochlear Implant Devices for Effective Speech Perception of Tonal Languages
(人工耳蝸中有效增強聲調語言感知能力的信號處理策略研究)
Principal Investigator: Tan Lee, Dept. of EE, CUHK
Co-Investigators: Andrew C. van Hasselt, Dept of Otorhinolaryngology, Head and Neck Surgery, CUHK
Sigfrid D. Soli, House Ear Institute, USA
Michael C.F. Tong, Dept of Otorhinolaryngology, Head and Neck Surgery, CUHK
Research Staff: Meng Yuan Dept. of EE, CUHK
Research Students: Yujia Li Dept. of EE, CUHK
Feng Huang Dept. of EE, CUHK
Hearing loss is one of the most common disorders in human beings. In Hong Kong, about 165 infants were born with severe hearing impairment every year (at an annual birth rate of 55,000). In China, there are over 27 million hearing-impaired people among whom over 3 million have severe hearing loss. Cochlear implant (CI) is a surgically implanted electronic device that provides a sense of sound to a person who is profoundly deaf. It delivers electrical stimulations directly to auditory nerves in the cochlea. From 1995 to 2007, over 4,000 CI surgeries were completed in China. In the coming few years, China will be having at least 15,000 more pediatric CI surgeries.
At present, nearly all commercial CI products are developed and manufactured by one of the three companies from U.S.A., Australia and Europe. Existing CI technologies have been developed and evaluated predominantly in western countries. They are not optimized for Chinese users who speak tonal languages like Mandarin and Cantonese. In Chinese dialects, lexical tone plays an important role in speech perception and understanding. A word carrying different pitch patterns may have totally different meanings. It is believed that tone-related speech cues need to be better encoded in CIs to achieve maximum benefit to the huge population of Chinese-speaking CI patients.
Research on CI processing strategies typically starts with acoustic simulation with normal-hearing (NH) subjects. In this approach, sound signal is processed in such a way that the acoustic output, when perceived by NH listeners, sounds similar to the electric output perceived by CI users. It is an effective way to attain a good understanding about the perceptual effects of a new signal processing algorithm before proceeding to real CI tests. Based on the findings of acoustic simulation, CI processing strategies are designed, implemented and tested on patients. Recently a number of research platforms have been made available to facilitate flexible programming and tuning of signal processing algorithms on the most popular CI products.
Our proposed research aims to develop tailor-made CI processing strategies for hearing-impaired tonal-language speakers. Tone-enhanced signal processing algorithms will be realized on state-of-the-art CI hardware and their effectiveness will be validated on real users. The population of CI users in China is growing rapidly in recent years. Our research results will directly benefit these users and improve the quality of their life. We expect that our methods will also be helpful improving music perception of general CI users.