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

作　　者：杨丹[1] 徐彬[1] 李封[1] 王旭[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110819

出　　处：《东北大学学报（自然科学版）》2014年第2期212-216,共5页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(60970157);中央高校基本科研业务费青年教师科研启动基金资助项目(N100304008)

摘　　要：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）.

关 键 词：语音编码策略 Gammatone滤波器 Meddis内毛细胞模型 心理声学 掩蔽效应 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]