【摘要】 在传统人工耳蜗连续交叠采样(Continuous Interleaved Sampler,CIS)算法的基础上,提出一种基于精细结构(频率调制信息)的人工耳蜗语音处理算法,在不引入过高频率成分、保证工艺可实现性的前提下,使语音识别率大幅提高。听觉仿真实验的结果表明,与传统的基于时域包络的CIS算法相比,基于精细结构的CIS算法对于元音可懂度的改进可以达到28%;声调的识别率在各种噪声条件下提高20%以上;在一般噪声环境下,辅音和句子的可懂度也分别获得了22.9%和28.3%的改进。更多还原

【Abstract】 A strategy was proposed to improve the performance of continuous interleaved sampling algorithm by in- troducing partial temporal fine structure cues,namely frequency modulation (FM) information,into the slowly varying temporal envelops.The improved algorithm has its own application values because it does not introduce too much high-frequency components into the model,which can not be perceived by deaf patients.The psychoacoustic experimen- tal rtsults show that the introduction of the frequency modulation information can improve the cochlear implant (CI) performance greatly:the intelligibility of vowel perception can get over 28% improvements,tone recognition scores can increase more than 20% at various noise conditions,and under moderate noise conditions the consonant and sentence recognition can also be improved 22.9% and 28.3% respectively.更多还原