基于分段块频域自适应滤波算法的长延时回声消除

在网络电话、电话会议、车载电话、IP电话等免提电话系统以及其他一些通信系统中,要想得到高质量的回声消除效果,必须抑制声学回声、尤其是长延时回声.然而对于长延时回声的抑制,如果采用传统的基于LMS/NLMS的自适应算法,则会产生大而复杂的计算量,且收敛速度慢.为解决此问题,本文对分段块频域自适应滤波算法进行了研究,即首先对该算法进行了描述,然后通过录制两个语音信号进行仿真.仿真结果表明,该方法具有非常好的长延时声学回声消除效果.

Long Time Delay Echo Cancellation Algorithm Based on Partitioned Block Frequency Domain Adaptive Filter Algorithm

In hand-free systems,such as vocie over internet protocal(VOIP),voice conference,car-kit and IP,and ofler telecommunication systems,it is necessary to suppress echo,especially the long time delay echo,to obtain a high quality of acoustic echo cancellation performance.If using the classical LMS/NLMS-based adaptive filter algorithm to cancell the long time delay echo,it will bring a large and complex calculation amount,and the convergence is slow.To solve this problem,the partitioned block frequency domain adaptive filter algorithm(PBFDAF) is presented.It is simulated by two recorded speeches in the real environment.Simulation results show that the algorithm has a good acoustic echo cancellation effect of long time delay.