三阶累积量的语音激活检测方法  

Speech signal activity detection algorithm based on third order cumulant

作　　者：刘福星[1] 何选森[1] 

机构地区：[1]湖南大学计算机与通信学院,长沙410082

出　　处：《计算机工程与应用》2011年第17期137-139,共3页Computer Engineering and Applications

基　　金：湖南省湘潭市科技计划项目(No.ZJ20071008)

摘　　要：在电子与通信系统中,传输信道的噪声都可以看作是加性的高斯随机过程,而高斯随机过程的三阶累积量为零,通信系统中传输的语音信号一般是非高斯信号,基于这个原理提出一种语音激活检测算法。在通信系统的接收端,对于被噪声污染了的语音信号,通过计算接收信号的三阶累积量,得到语音信号的累积量,从而可以区分语音和噪声,达到检测出语音信号的目的。仿真结果表明,在通信系统低信噪比的环境下能有效地检测出语音信号。In communication system,most channel noise can be considered as Gaussian random processes and the third order cumulant of Gaussian random signal is zero.The speech signal is not Gaussian random signal in general.This paper presents a novel speech signal activity detection algorithm based on the third order cumulant of noisy speech signal equaling to the cumulant of speech signal, so speech and noise can be distinguished by using the third order cumulant.The results of simulation show that this algorithm can effectively detect speech signal under low signal to noise ratio condition.

关 键 词：三阶累积量 语音激活检测 高斯噪声 

分 类 号：TN911[电子电信—通信与信息系统]