非线性混合模式的语音盲分离算法

本文针对FIR非线性混合模型，基于最大熵算法，提出了一种以高斯混合模式概率密度函数估计替代传统对数化概率密度估计的盲分离算法，以偶函数为非线性激活函数，采用最大期望（EM）迭代算法推导了分离算法的权向量迭代公式，通过模拟仿真实验结果与传统的最大熵和高阶累积量方法比较，新算法提高了收敛速度，并有效地完成了非线性语音分离任务，抑制了干扰语音信号的影响，提高了输出信噪比。

Blind source separation for nonlinear speech mixture

Based on the FIR nonlinear mixture model and the Maximizing Entropy principle, another speech separation batch processing algorithm is proposed by using the Gaussian Mixture Model probability density function (PDF) estimation in stead of the logarithm PDF estimation. On applying even nonlinear function, an iterative algorithm based on Expectation-Maximization is provided. Computer simulation shows that this algorithm can separate source effectively. Based on the comparison of the different algorithms, It can be concluded that the proposed separation algorithms has good convergence and is robust. The outputs' signal-noise-ratios are improved.