电子耳蜗n of m算法的特性

摘要

目前主流的电子耳蜗言语处理策略包括基于提取特征的算法和基于滤波器组的算法.SPEAK算法和ACE算法是主要的基于滤波器组的算法策略,而从算法结构上来说,都属于n of m策略.n of m策略的频带划分数量和最大谱峰选择数量可调,有助于基于语音特征更好地传递目标信号.本文研究电子耳蜗n of m算法的特征,比较算法参数特征的变化,为算法改进及算法嵌入提供研究基础,具有较高的理论价值和应用价值.%At present,the mainstream speech processing strategy of cochlear implant includes feature extraction-based algorithm and filter bank-based algorithm.SPEAK and ACE algorithms are the main filter bank-based algorithms,which can be considered to be n of m strategy as their algorithm structures.The n of m strategy has adjustable quantity of frequency bands and maximal spectral peak,which helps to transfer desired signal based on speech features.Herein,the features of n of m algorithm for cochlear implant were studied and the changes of algorithm parameter features were compared.With theoretical and practical values,the study provides research basis for algorithm improvement and algorithm implanting.