基于听觉特性的电子耳蜗语音编码策略

Auditory Characteristics-based Speech Coding Strategy in Cochlear Implant

为了改善电子耳蜗语音编码策略的性能,提出了一种基于听觉特性的编码策略.策略中的带通滤波及包络检测阶段采用Gammatone滤波器模型和Meddis内毛细胞模型;在通道选择阶段引入心理声学的掩蔽效应模型,从M个通道中选出N个通道,在相应的通道产生脉冲刺激信号.用MATLAB对提出的策略进行仿真研究,实现了22选8的编码策略.结果表明,所提编码策略在稳定性和鲁棒性方面优于CIS和ACE编码策略.

For the limitations of speech coding strategy in cochlear implants, a coding strategy was proposed based on auditory characteristics. For the band-pass filter and envelope detector stage in the proposed strategy, Gammatone filterbanks model and Meddis inner hair cell （IHC） model were used respectively. For channel selection stage, a psychoacoustic masking effects model was used. The strategy using MATLAB was simulated, and 8-of-22 channel coding strategy was realized. The results showed the proposed strategy had more stability and robustness than other strategies, such as CIS （continuous interleaved sampling）and ACE （advanced combination encoder）.