一种语音信号端点检测的改进方法  

An Improved Method of the Voice Endpoint Detection

作　　者：顾亚强[1] 赵晖[1] 吴波[1] 

机构地区：[1]国防科学技术大学电子科学与工程学院二系,湖南长沙410073

出　　处：《计算机仿真》2010年第5期340-343,共4页Computer Simulation

基　　金："十一五"武器装备预研项目(51329060101)

摘　　要：在研究传统的语音端点检测方法的基础上,为提高语言识别率和准确性,提出了一种新的利用对短时能零和过零率的进行差分的方法来求取语音的起始点,达到了语音端点检测的目的。利用语音段和无声段的边界差异性,运用差分的方法找到语音段的起点和末点,避免了对语音段本身的判定过程,增强了抗噪性能也降低了计算复杂度。通过仿真实验结果可以证明,上述方法优于传统语音端点检测方法。A new method of voice endpoint detection using the difference of the vocie short energy and cross zero ratio is proposed on the basic of the research of the traditional method. Since the difference between the speech segment and silence segment, the start point and end point can be found with this method by using difference. It ＇ does not have to deal with the voice segment itself. It improves the performance in the noise environments and can reduce the computation. It is proved that this new method is better than the traditional method through the experiment.

关 键 词：语音端点检测 短时能零 过零率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]