一种会议电话回波抵消自适应算法

New Adaptive Algorithm on Echo Cancellation of Teleconference

最小均方自适应算法是符合国际电信联盟(ITU)标准G.168的回波抵消自适应算法.然而,单纯符合该标准并不能保证完全满足消费者所期望的典型的话音质量,并且不适用于电话会议的回波抵消系统.为克服这个缺点,本文提出了一种适用于电话会议的回波抵消算法:归一化最小均方(NLMS)自适应算法.这种算法的收敛条件与输入信号特征无关,同时,当输入信号为语音时,NLMS自适应算法的收敛速率更快,稳定性好.仿真实验表明:NLMS算法优于LMS算法.

LMS algorithm is a kind of adaptive algorithm for echo cancellation which accords with ITU （ International Telecommunications Union ） standard. But the simple accordance with the standard it cant promise to satisfy the typical acoustic quality anticipated by the consumers , and it doesnt accord with the echo cancellation system of the telephone conference To overcome the disadvantage, the paper proposes a kind of echo cancellation algorithm which is called the adaptive algorithm of normalized least mean squares（NLMS） for teleconference. The convergence condition of the algorithm is independent of input signals＇ characteristics, and, when the input signal is voice signal, the convergence rate of NLMS adaptive algorithm is higher than that of LMS adaptive algorithm, the stability of NLMS is better than that of LMS. The result of the simulation experiment shows that NLMS algorithm is better than LMS algorithm.