A permutation alignment algorithm based on multiple criterions in audio signal Frequency-Domain Blind deconvolution

In audio signal frequency-domain blind deconvolution, the conversion of signals from time-domain convolutive mixture to several instantaneous mixtures in frequency-domain may cause indeterminacies in scaling and permutation. Based on traditional permutation alignment algorithms, this paper proposes a permutation alignment algorithm based on multiple criterions. A combination of several classical criterions will be used to make a more precise permutation judgment. Simulation results show that algorithm based on multiple criterions achieves a better accuracy in permutation alignment and the acoustic quality of separation is improved.