An improved speech coding strategy for cochlear implants

Abstract

Cochlear implants are widely accepted as the unique and most effective ways for individuals with severe to profound hearing loss to restore some degree of hearing. Currently it is difficult for cochlear implants to transmit more frequency and phase information, which are crucial for the perception of tonal language and of hearing in realistic listening conditions. An speech coding strategy based on improved one-octave wavelet transform zero-crossings stimulation(IWZCS) is proposed to encode more frequency and phase information in a certain way, which may more appropriable for cochlear implant users. IWZCS generates stimulus in the domain of wavelet transform in lower frequency bands and synthesizes stimulating pulsatiles using fundamental frequency (F0) modulation on the basis of wavelet domain zero-crossings for higher frequency bands. With amplitude modulation and more frequency and phase information (zero-crossings) encoded, the IWZCS is aimed at improving the recognition of tonal language and speech in noisy environment. CIS, FAME and IWZCS are compared through computer simulation and results show that correlations with original speech signal is significantly higher in signal synthesized through IWZCS than in signals reconstructed by CIS and FAME strategies. Based on spectrum analysis, IWZCS strategy may allow better recognition in tonal language and better speech understanding in noise.