一种基于ZSP800核改进型LMS自适应滤波算法的实现

摘要

针对手机终端语音降噪处理系统中，广泛使用计算复杂度较小的固定步长LMS（Least Mean Square）自适应滤波算法。本文提出一种通过建立步长因子µ(n)和误差信号e(n)非线性函数关系，进而把固定步长LMS自适应算法改进为新的变步长自适应滤波算法，较好地解决了固定步长因子无法解决收敛速度和稳态误差之间相矛盾的不足。最后，给出了改进型算法在ZSP800核数字信号处理器（DSP）上的实现方法。%Aiming at the denoising system of mobile terminators, with the wide use of steady step size LMS adaptive filtering algorithm. This paper proposes a way by building a nonlinear function relationship between step size and the error signal, thus we change the SSLMS algorithm to variant step size LMS algorithm. It solves the convergence speed contradicting its steady-state error in SSLMS algorithm perfectly. At last, the paper gives the implementation of the modiifed LMS algorithm based on the ZSP800 core digital signal processor.