Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (34): 55-56.

• 学术探讨 • Previous Articles     Next Articles

Speech endpoint detection method based on signal subspace and information complexity

TAN Qiao-lai1,QIAN Sheng-you1,CHEN Ya-qi2   

  1. 1.School of Physics and Information Science,Hunan Normal University,Changsha 410081,China
    2.Department of Physics,Xiangnan University,Chenzhou,Hunan 423000,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-12-01 Published:2007-12-01
  • Contact: TAN Qiao-lai

基于信号子空间和信息复杂度的语音端点检测

谭乔来1,钱盛友1,陈亚琦2   

  1. 1.湖南师范大学 物理与信息科学学院,长沙 410081
    2.湘南学院 物理系,湖南 郴州 423000
  • 通讯作者: 谭乔来

Abstract: A new speech endpoint detection method is proposed based on speech enhancement with signal subspace approach and information complexity in additive noise environment.In this method,the additive noise is removed using signal subspace approach at first,then the information complexity based method is used to detect the endpoints for the enhanced speech.Experimental simulations show that this method can increase the endpoint detection accuracy ratio compared with the traditional speech endpoint detection method,especially it has high endpoint detection accuracy ratio under low signal-to-noise ratio.

Key words: speech endpoint detection, information complexity, signal subspace

摘要: 针对染有加性噪声的语音信号,提出了一种基于信号子空间和信息复杂度相结合的语音端点检测方法。该方法先利用信号子空间法去除加性噪声,再对增强后的语音利用信息复杂度进行端点检测。实验仿真表明,该方法相对传统的语音端点检测方法,能提高语音端点检测准确率,特别在低信噪比条件下具有较高的端点检测准确率。

关键词: 语音端点检测, 信息复杂度, 信号子空间