用于回声消除系统的自适应延时估计算法研究

为避免回声消除系统中滤波器过长导致的各种缺陷,提出一种单滤波器结构的自适应延时估计(delay estimation,DE)算法,并与现有的两种双滤波器延时估计算法S-DE、PHT-DE进行对比研究。先阐述自适应延时估计算法的原理,并对三种算法进行理论研究,然后采用高斯白噪声作为输入信号,以稀疏回声路径为实验条件,对算法进行计算机仿真。仿真结果表明,与PHT-DE相比,S-DE算法计算复杂度较低而稳定性较弱,但两种算法均为双滤波器结构,因此存在信息冗余而导致计算复杂度浪费。新算法为单滤波器结构,可有效避免双滤波器结构的信息冗余,其性能与计算复杂度均优于S-DE、PHTDE,是各种DE算法里实用性最强的算法。最后指出该算法的改进方向是进行统计学建模,为参数设置提供理论指导公式。

RESEARCH ON ADAPTIVE DELAY ESTIMATION ALGORITHMS USED FOR ECHO CANCELLATION SYSTEM

In order to eliminate the disadvantages resulted from excess length of adaptive filter used for echo cancellation system, an adaptive delay estimation (DE) algorithm with single filter structure is proposed and then compared with the two existing dual filter delay estimation algorithms: S-DE algorithm and PHT-DE algorithm. The theory of delay estimation is described and three delay estimation algorithms above are studied theoretically, and then simulated with Gaussian input and sparse echo path. Simulation results demonstrate that compared with PHT-DE, S-DE has lower computational complexity and weaker robustness. However both algorithms have information redundancy due to their dual filter structure, which leads to the waste of complexity. The proposed algorithm which owns single filter structure eliminates the information redundancy of dual filter structure and it is superior to other two algorithms in both performance and computational complexity. Consequently, it should be the most practicable algorithm. Finally, it comes to the conclusion that the proposed algorithm should be improved by setting statistical model to provide theoretical direction for parameter settings.