基于张量分解的卷积盲源分离方法  

Convolutive blind source separation method based on tensor decomposition

作　　者：马宝泽 张天骐[1] 安泽亮 邓盼 MA Baoze;ZHANG Tianqi;AN Zeliang;DENG Pan(Chongqing Key Laboratory of Signal and Information Processing,School of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]重庆邮电大学通信与信息工程学院信号与信息处理重庆市重点实验室,重庆400065

出　　处：《通信学报》2021年第8期52-60,共9页Journal on Communications

基　　金：国家自然科学基金资助项目(No.61671095,No.61371164);信号与信息处理重庆市市级重点实验室建设基金资助项目(No.CSTC2009CA2003);重庆市教育委员会科研基金资助项目(No.KJ130524,No.KJ1600427,No.KJ1600429)。

摘　　要：基于张量分解框架提出了一种卷积盲源分离方法,同时解决了混合滤波器矩阵估计和频点排序的问题。首先,根据观测信号的估计自相关矩阵构造出所有频点处的张量模型;然后,利用张量分解技术计算出每个频点上对应的因子矩阵作为该频点的估计混合滤波器矩阵;最后,采用以功率比作为测度的全局优化排序策略消除了全频段的排序模糊性。实验表明,所提方法在不同仿真条件下处理卷积混合的实测语音时表现出了比现有算法更优异的分离性能。A convolutive blind source separation algorithm was proposed based on tensor decomposition framework,to address the estimation of mixed filter matrix and the permutation alignment of frequency bin simultaneously.Firstly,the tensor models at all frequency bins were constructed according to the estimated autocorrelation matrix of the observed signals.Secondly,the factor matrix corresponding to each frequency bin was calculated by tensor decomposition tech-nique as the estimated mixed filter matrix for that bin.Finally,a global optimal permutation strategy with power ratio as the permutation alignment measure was adopted to eliminate the permutation ambiguity in all the frequency bins.Expe-rimental results demonstrate that the proposed method achieves better separation performance than other existing algo-rithms when dealing with convolutive mixed speech under different simulation conditions.

关 键 词：卷积盲源分离 张量分解 自相关矩阵 排序模糊 

分 类 号：TN911.6[电子电信—通信与信息系统]