A Hybrid Coding Strategy to Improve Auditory Perception of Cochlear Implant

摘要

Sensory deafness patients can resume the hearing by cochlear implant (CI), but in noisy environment, auditory perception is seriously in decline, in addition, most speech coding techniques of CI are amplitude coding for no tone language, lacking the extraction of information for time fine structures. This paper proposes the hybrid coding strategy (HCS) to improve the auditory perception of tone language. Against the noise problem, a noise removing algorithm based on auditory masking properties is proposed to filter noise in sub-band. For low frequency channels, a model of the peripheral auditory system is used to generate the stochastic impulse, which is closer to the physiological activity of auditory neural firing. For high frequency channels, a frequency shifting method is used to shift the information of high frequency channels to low frequency channels, which can make patients acquire more speech message. Compared with previous methods, the fundamental frequency of proposed strategy is closer to actual fundamental frequency, and under the same condition of SNR, the recognition rate of Mandarin tones, words, and sentences is also significantly better than that of Continuous interleaved sampling CIS, Advanced Combination Encoding(ACE), and Phase Locking Stimulating (PLS) strategies.