摘要
使用听觉感知的小波变换来提取电子耳蜗中的共振峰参数。首先用听觉感知的小波变换对原始语音信号进行分解重构,然后分别用自相关和格型法对合成语音信号和原始语音信号进行共振峰提取。实验结果表明:使用听觉感知的小波变换进行共振峰参数提取的可行性,合成语音信号能更好地表征原始语音信号的特征;同时也证实了电子耳蜗语音处理器中使用由格型法提取共振峰参数比自相关法更精确。
Extract cochlear formant on the basis first,then extract the formant of synthetic speech of auditory perception.Filter the speech and original speech with self-correlation with a set of auditory perception filter algorithm and grid model algorithm.The experiment results show that extracting cochlear formant on the basis of auditory perception is viable,the character of synthetic speech signal achieved in this paper could token original speech＇s character and the grid model algorithm is stabilizing and exactitude than self-correlation algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第29期232-234,共3页
Computer Engineering and Applications
基金
国家自然科学基金( the National Natural Science Foundation of China under Grant No.60572076)
江苏省高校自然科学研究计划项目资助( 05JKB510113)
关键词
共振峰提取
听觉感知
电子耳蜗
格型法
auditory perception wavelet transform
formant extraction
cochlear
grid model algorithm
