Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (27): 114-116.DOI: 10.3778/j.issn.1002-8331.2010.27.031

• 数据库、信号与信息处理 • Previous Articles     Next Articles

Chinese continuous speech endpoint detection method based on vowel detection

ZHANG Li-ping,FENG Hong-wei,WANG Yan   

  1. College of Information Science and Technology,Northwest University,Xi’an 710127,China
  • Received:2009-06-23 Revised:2009-09-16 Online:2010-09-21 Published:2010-09-21
  • Contact: ZHANG Li-ping

基于元音检测的汉语连续语音端点检测方法

张利平,冯宏伟,王 艳   

  1. 西北大学 信息科学与技术学院,西安 710127
  • 通讯作者: 张利平

Abstract: Accurate detection for the start and end point of syllable is a very important step in Chinese continuous speech recognition.Traditional endpoint detection methods achieve high rate of success in non-continuous speech,but fall in continuous speech.In this paper,a new method based on MFCC0 and the formant energy of Chinese vowels is proposed,which can accurately detect the start and end point of syllable.The results of experiments show that this method achieve high rate of success even in low SNR.

摘要: 在汉语连续语音识别中,准确检测出音节的始点和终点是很重要的一步,传统的端点检测方法在非连续语音中检测准确度很高,但在连续语音中检测准确度会大幅度降低。利用MFCC0参数和汉语元音的共振峰能量设计了一种新的端点检测方法,可以准确检测出汉语连续语音中的音节端点。实验结果表明:这种端点检测方法在低信噪比下也有很高的检测正确率。

CLC Number: