基于小波域的非平稳卷积混合语音信号的自适应盲分离

Adaptive wavelet domain algorithm for blind separation of non-stationary convolved speech mixtures

一些卷积混合信号的盲分离算法是迭代性的,不适于实时应用.为此提出一种基于小波域的算法,用于卷积混合信号的自适应盲分离.对基于小波域的算法进行仿真,并与频域盲信号分离算法进行对比,结果表明所提出的算法能提高盲信号分离的性能.

A divide-and-conquer strategy is implemented by using the wavelet transform. Based on the wavelet domain algorithm, blind separation of non-stationary convolved speech mixtures is performed. And better performance is obtained compared with the frequency domain algorithm.