基于小波变换的两路自适应回声消除算法

Research on Two-Channel Acoustic Echo Cancellation Using Adaptive Filtering Algorithm Based on Wavelet Transform

免提式语音通信系统在VoIP(Voice over Internet Protocol),手持移动终端和远程会议系统应用得非常广泛.但是,由于扬声器和麦克风之间的耦合而产生的声学回声,严重影响了语音的清晰度.常用的NLMS2算法在遇到相关性很强的两路语音信号时,收敛速率变得很慢,为改进此算法,基于小波变换和语音信号能量分布的特点,提出了一种新的两路自适应回声消除算法——WL-APN-LMS2.计算机仿真结果表明,该算法比传统的NLMS2、AP2算法收敛性能更好,计算复杂度较低,综合性能更优.

Hands-free voice communication system has a wide range of applications to VoIP,handheld mobile terminals and remote conferencing systems.However,as the speakers and microphone caused acoustic echo because of the coupling,the clarity of voice gets a serious impact.The convergence rate of NLMS2 algorithm will get slow when the two-way voice signals are highly relevant;in order to improve this algorithm,we propose a new algorithm named WL-APNLMS2 based on wavelet transform and the energy distribution of speech signals in two-channel adaptive echo cancellation.The computer simulation results show that the new algorithm has better convergence than the traditional algorithm NLMS2,AP2;and the computational complexity is also low.