用于回波抵消的最大长度序列相关近端语音检测算法研究

Research of Maximum Length Correlation Near - end Voice Detection Algorithm in Acoustic Echo Canceling

回波抵消是语音通信系统中不可缺少的一个重要组成部分。大部分回波抵消技术都是基于自适应LMS算法的。在实际应用中,近端语音检测的准确性会在很大程度上影响自适应LMS算法在双向通信环境下回波抵消的效果。本文提出了一种基于最大长度序列相关算法的近端语音检测算法。这种近端语音检测算法和自适应LMS算法相结合,得到的回波抵消算法在模拟双向通信环境下的回波抵消效果比自适应LMS算法高约8db)。

Acoustic echo canceling is an essential portion of speech communication systems. Most of which arc based on self - adapting LMS algorithm. In applications, the accuracy of the near - end voice detection will severely affect the performance result of the self- adapting LMS algorithm in two- way communication. In this paper, we suggest a new near -end voice detection method, which is based on maximum length correlation algorithm. Combining this detector with self- adapting LMS algorithm, we get a new acoustic echo canceling algorithm which can provide ERLE about 8db more than that of pure self - adapting LMS algorithm for two - way voice transmission.