采用短时傅立叶变换方法的电子耳蜗语音处理技术  

Performance of STFT in speech processing used for cochlear implant

作　　者：陈红芳[1] 冯海泓[1] 徐海东[1] 

机构地区：[1]中国科学院声学研究所东海研究站,上海200032

出　　处：《声学技术》2007年第3期442-446,共5页Technical Acoustics

摘　　要：电子耳蜗是用有限个电极刺激听神经以恢复全聋人听觉的装置,目前发展的关键目标是研制出可靠、灵活、价格低廉的产品。在介绍了电子耳蜗的原理和语音信号处理方案的发展的基础上,研究了采用短时傅立叶变换方法的电子耳蜗语音处理技术。据此法分析提取参数用以刺激耳蜗内的听神经,使聋人恢复听觉,该技术的优点在于:能根据每个耳聋病人的毛细胞和神经残留多少等情况的不同来灵活的选择刺激参数,达到较好的语音识别效果。Electronic cochlear implant is an electronic device for restoring deaf people＇s hearing with limited electrodes implanted in their cochlear. The purpose of the present work is to develop a flexible product with high reliability and reasonable price. The principles of the cochlear implant and the speech processing strategies are presented. Performance of STFT method in speech processing is proposed for cochlear implant. This method is used for picking up parameters properly to stimulate the auditory nerve of a totally or seriously deaf person, and has great flexibility to choose basic speech characteristics being considered or to adapt special stimulation parameters for a specific patient to get better speech recognition.

关 键 词：电子耳蜗 短时傅立叶变换(STFT) 语音信号处理 语音识别 

分 类 号：TN911.7[电子电信—通信与信息系统]