语音端点检测和基于自适应控制的语音噪声滤波  

Speech Endpoint Detection and Speech with Noise Filter Based on Adaptive Control

作　　者：武毅博 李霏[1] 黎文[1] 

机构地区：[1]武汉理工大学自动化学院,湖北武汉430070

出　　处：《工业控制计算机》2015年第8期93-95,共3页Industrial Control Computer

摘　　要：介绍了基于自适应控制的噪声滤波抵消系统,提出一种易于实现的LMS算法。通过Matlab仿真实验,采用自适应陷波器明显改善了带有噪音的语音信号的信噪比。在语音信号端点检测中,短时倒谱距离值作为特征参数。采用双门限检测法来判断语音帧和非语音帧。通过仿真表明该方法可较为准确地检测到语音信号的端点。This paper introduces the adaptive nosie system which based on the basic principle of adaptive control.A LMS algorithm that is easy to achieved is proposed.The Matlab simulation result shows that the adaptive notch filter can improve the SNR of the speech signal with noise obviously.Cepstral distance as a characteristic parameter in the voice endpoint detection.Using the double threshold method,the speech frame and the non speech frame can be distinguished.

关 键 词：LMS算法 倒谱距离 端点检测 MATLAB仿真 

分 类 号：TN912.3[电子电信—通信与信息系统]