利用声回波对消器输出信号构建检测统计量的双端语音检测
Doubletalk Detection Using Detection Statistics Formed by the Output Signal of the Echo Canceller
作者：徐自励(四川大学 电子信息学院,四川 成都610064)；何培宇(四川大学 电子信息学院,四川 成都610064)；周激流(四川大学 电子信息学院,四川 成都610064)
Author：(School of Electronic and Info.,Sichuan Univ.,Chengdu 610064,China)；(School of Electronic and Info.,Sichuan Univ.,Chengdu 610064,China)；(School of Electronic and Info.,Sichuan Univ.,Chengdu 610064,China)
收稿日期：2005-06-09 年卷（期）页码：2006,38(1):133-135
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：双端语音检测;互相关;声回波对消;自适应滤波
Key words：doubletalk detection(DTD);cross-correlation;acoustic echo cancellation(AEC);adaptive filter
基金项目：国家自然科学基金资助项目(60472096)
中文摘要
声回波对消中，双端语音检测用以判定远端语音是否混和有近端语音。为了提高双端语音检测性能，采用自适应声回波对消器输出信号(而非远端语音信号)与麦克风输出信号构建检测统计量，利用无近端语音时二者间较强的相关性，达到提高检测性能，减少运算量的目的。通过算法软件仿真，结果证明该算法较其它的互相关统计量检测算法具有较好的检测性能。
英文摘要
Doubletalk detection (DTD) is used to sense whether the far-end speech is corrupted by the near-end speech in echo cancellation. In order to improve the performance of doubletalk detection, the output of an adaptive acoustic echo canceller (AEC), instead of a far end speech, and the output of a microphone were employed to form detection statistics. By taking advantage of the strong correlativity between the output of the AEC and the output of the microphone when the near-end speech is absent, the detection performance was improved and the computational complexity was decreased. Simulation results showed that better performance of this algorithm compared with the performance of other cross-correlation based algorithms can be achieved.
【关闭】
