强噪声环境下改进的语音端点检测算法  

Improved speech endpoint detection algorithm in strong noise environment

作　　者：鲁远耀[1] 周妮[1] 肖珂[1] 叶青[1] 

机构地区：[1]北方工业大学信息工程学院,北京100144

出　　处：《计算机应用》2014年第5期1386-1390,共5页journal of Computer Applications

基　　金："十一五"国家科技支撑计划重点项目(2009BAI71B02)

摘　　要：为了提高强噪声环境下语音端点检测的正确率,克服传统的短时能量和短时过零率双门限语音端点检测算法在低信噪比(SNR)条件下检测性能急剧下降这一缺陷,提出了一种改进的语音端点检测算法。该方法对强噪声环境下的语音信号,首先进行小波阈值去噪,提高信噪比,再采用双门限法进行端点检测。实验结果表明,该算法具有一定的鲁棒性,在强噪声环境下仍能准确地进行语音端点检测,从而该算法的有效性得到验证。To improve the correctness of speech endpoint detection in strong noise environment, and overcome the disadvantage in traditional dual-threshold speech endpoint detection based on short-time energy and short-time zero-crossing rate, whose performance degrades sharply in low Signal-to-Noise Ratio （SNR） environment, an improved speech endpoint detection algorithm was proposed. In this method, the noisy speech signal was de-noised to enhance SNR at first, then the dual-threshold speech endpoint detection algorithm was used to detect the endpoints of the de-noised speech signal. The experimental results indicate that the proposed method not only has strong robustness, but also can achieve high detection accuracy in strong noise environment, so the effectiveness of the algorithm is proved.

关 键 词：端点检测 小波阈值 去噪 双门限 信噪比 

分 类 号：TN912.34[电子电信—通信与信息系统]