一种新的语音噪声抵消算法

针对单参考输入自适应噪声抵消器在多噪声源情况下除噪声效果不理想的问题，根据多参考输入自适应噪声抵消器模型，在推导多参考输入最小均方（MLMS）算法的基础上，给出其改进算法（IMLMS），结合语音信号的特点，给出了多参考输入自适应语音消噪算法（VMLMS），利用调整算法执行过程的算法改进思想，突破了传统的修正收敛系数的算法改进模式，最后对几种算法的性能进行了计算机仿真比较。

A new algorithm of eliminating speech noises

Normally, it is not ideal for single reference adaptive noise canceller to eliminate noises under circumstance of more than one noise source. So according to the model off multiple reference adaptive noise canceller, this paper deduces a multiple reference least mean square algorithm (MLMS) and presents its improved algorithm (IMLMS). This improved algorithm has some advantages, such as fast convergence and doing well in eliminating noises under circumstance of more than one noise source. As an application in voice signal processing, this paper presents an multiple reference adaptive algorithm by adjusting algorithm process compare with the simulation results. According to the analysis of this algorithm and simulation results, it could be concluded that multiple reference adaptive algorithm of eliminating speech noises has the advantages of simplicity, stability, and easy to implement.