基于相邻频点幅度相关的语音信号盲源分离

Blind source separation of speech signals based on the amplitude correlation of neighbour bins

排序和幅度不一致性是在频域进行信号盲源分离的主要困难。针对语音信号邻近频点间信号幅度相关性能良好这一特点,本文提出基于相邻频点间幅度相关的盲源分离算法,用以消除卷积信号盲源分离过程中排序不确定性。本算法理论简单,稳健性好。仿真结果表明该方法对卷积混合后的语音信号能得到较好的分离效果,并且耗时较短。

Indeterminacies in amplitude and permutation are the two main cumbersome aspects for blind signal separation in frequency domain. Based on the fact that there is a good amplitude correlation between neighbour bins of a speech signal, we presented a new blind source separation method that can eliminate the permutation indeterminacy. Since there is no complicated mathematic derivation in this method, it is relatively simple in theory. Furthermore, the outcome of multiply simulation experiments confirmed its validity in blind source separation of speech signals. In addition, this method is robust and time saving.