一种基于信息熵的语音端点检测方法

摘要

A new scheme of speech endpoint detection based on informationentropy was presented in this paper . We exploited the properties of the speech and the information entropy function, and constructed an entropy function adapted to speech signal processing. Experiments showed that the new scheme could obtain high scores on speech endpoint detection,and its complexity was very low.%根据语音信号的波形特征，利用熵函数的性质，构造了一种特殊的熵函数，通过判断此函数值的大小，确定是语音还是无声状态。实验表明，此方法计算简单而且具有很高的准确性。