基于改进语音信号双门限法端点检测算法的研究  

Research on speech signal endpoint detection algorithm based on improved double threshold method

作　　者：朱敏[1] Zhu Min(College of Electronic and Electrical Engineering,Anhui Sanlian University,Hefei 230601,China)

机构地区：[1]安徽三联学院电子电气工程学院,安徽合肥230601

出　　处：《湖南文理学院学报（自然科学版）》2023年第3期48-53,共6页Journal of Hunan University of Arts and Science(Science and Technology)

基　　金：安徽省自然科学重点研究项目(KJ2021A1190,2022AH052002)。

摘　　要：语音信号端点检测是语音信号预处理过程中的重要环节,传统双门限法采用短时能量和短时平均过零率,通过设置阈值进行语音信号起始点和结束点检测,在高信噪比条件下识别效果较好,但是在低信噪比下,噪声影响传统检测的速度和准确性。为了提高语音识别的效率,提出一种改进的双门限法语音端点检测算法,采用动态设定阈值,进行平滑滤波等改进方法。通过MATLAB仿真表明,改进算法在低信噪比下的识别准确性较高,有利于后续语音识别的研究。Speech signal endpoint detection is a key link in the process of speech signal preprocessing.The traditional double threshold method uses the short-time energy and short-time zero crossing rate on average,and detects the starting point and end point of speech signals by setting the threshold.The recognition effect under the condition of high SNR is better than that of low SNR,under which noise can have an impact on the speed and accuracy of traditional detection.In order to improve the efficiency of speech recognition,this paper proposes an improved double threshold speech endpoint detection algorithm,which adopts dynamic threshold setting and smooth filtering.MATLAB simulation shows that the improved algorithm has high recognition accuracy under low signal-to-noise ratio,which is conducive to subsequent research on speech recognition.

关 键 词：端点检测 双门限法 动态阈值 平滑滤波 

分 类 号：TN912.3[电子电信—通信与信息系统]