SPEECH ENHANCEMENT USING β- DIVERGENCE BASED NMF WITH UPDATE BASES

In this paper, combination of statistical model based approach and Non-negative matrix factorization (NMF) based approach with on-line update of speech and noise bases for speech enhancement is proposed. Template based approaches are more robust and performs better to non-stationary noises compared to the statistical model based approaches. However, the template based approach is dependent on a priori information. Combining the approaches avoids the drawbacks of both. To improve the performance further, speech and noise bases are adapted simultaneously in NMF approach with the help of the estimated speech presence probability (SPP). The proposed approach yields better results than statistical based approach, NMF based approach and also combination of both approaches without on-line update in non-stationary noise environments.