Single channel blind source separation based on NMF and its application to speech enhancement

In this paper, an improved nonnegative matrix factorization (NMF) algorithm is proposed for single channel blind source separation and applied to speech enhancement. By adding time correlation item to objective function to constrain the time-varying gain coefficients of noise, it can achieve better effect of speech enhancement. We propose an efficient algorithm to optimize objective function with constraint item and effectively improve the estimation precision of the time-varying gain coefficients. Experiment results show that the proposed algorithm has more advantages compared with existing methods.