语音信号端点检测方法与展望  

A survey of endpoint detection methods for speech signal

作　　者：杨胜跃[1] 周宴宇[1] 黄深喜[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410075

出　　处：《信息技术》2005年第7期5-8,共4页Information Technology

基　　金：国家自然科学基金资助项目(69975003)

摘　　要：对语音信号端点检测的主要方法,如基于短时能量的方法、基于HMM的方法、基于自相关相似距离的方法等进行了深入研究。分析了这些方法的原理、优点与不足,并进行了比较。针对各种检测方法在应用中,尤其是应用于噪声环境中的局限性,指出了语音信号端点检测领域进一步研究的方向。The main speech signal endpoint detection methods, such as short time energy based scheme, HMM based scheme, related alike scheme and so on are investigated deeply. The analysis about the principle of these methods, their advantages and disadvantages, and the comparison between these methods are given. According to their limitations, especially when applied in noisy environment, the research direction about speech signal endpoint detection is pointed out.

关 键 词：语音信号 端点检测 噪声 

分 类 号：TP206.1[自动化与计算机技术—检测技术与自动化装置]