一种改进的基于倒谱特征的带噪语音端点检测方法  

An Improved Speech Endpoint Detection Method based on Cepstrum Distance

作　　者：沈红丽[1] 曾毓敏[1] 王鹏[1] 

机构地区：[1]南京师范大学物理科学与技术学院,江苏南京210097

出　　处：《通信技术》2009年第2期156-158,共3页Communications Technology

摘　　要：环境噪声是语音识别和说话人识别性能下降的原因之一,端点检测作为其关键技术之一,性能优劣在某种程度上决定了识别率的高低。文章提出一种改进的基于倒谱特征的带噪语音端点检测方法。在传统基于倒谱距离的算法基础上,该算法进一步综合利用短时过零率和短时能量作为最终判决的门限。实验结果表明,该方法计算效率较高,低信噪比下有较好的检测性能。One of the causes in reducing the capability of speech recognition（SR） systems is the noisy enviroment. As an important technology in SR, the accuracy of endpoint detection determines the SR rates in some degree. In this article, an improved method based on MFCC for endpoint detection is proposed. Short time energy and zero crossing rate are intergraded as final decision threshold on the basis of traditional endpoint detection method based on MFCC. The experiments show that the high computational efficiency and good accurate detection can be obtained under the condition of low SNR.

关 键 词：端点检测 倒谱距离 带噪语音 

分 类 号：TN912.3[电子电信—通信与信息系统]