【摘要】 卷积混合语音进行盲源分离时,不能直接应用独立分量分析(ICA)算法。采用一种新的卷积混合语音模型,对多通道混合语音使用近来提出的时域EFICA算法进行盲分离,然后利用聚类和重构算法来恢复源信号。通过真实语音实验表明,提出的算法能有效地分离混合语音信号。更多还原

【Abstract】 BBS(Blind Source Separation)of convolution-mixed audio can not be applied directly to ICA(Independent Component Analysis)algorithm.A new convolution-mixed mode of the speech signal is used.Mixed audio multi-channel are processed using the recently proposed time-domain EFICA,then algorithm of clustering and reconstruction is used to restore the signal source.The experiments show that the algorithm can effectively separate mixed speech signal by the really audio.更多还原