Approach of feature with confident weight for robust speech recognition

Robustness enhancement has become one of the research focuses of the acoustic speech recognition system. In recent works, the missing feature theory (MFT) has been proved as an available and considerable solution for robust speech recognition based on either ignoring or compensating the unreliable components of feature vectors corrupted mainly by the band-limited background noise. Because of MPA classifying in a binary way and dealing with the cepstral feature, this paper proposes three new approaches based on confidence analysis. The approach of feature with confident weight (AFCW) estimates the confidence of each feature component as its weight and describes the effect of noise in a more precise way. The other two approaches, SC(simple cepstral) and TC- (total cepstral) AFCW, can be regarded as an AFCW on cepstral domain. Experimental results have shown that the proposed approach could significantly improve the recognition accuracy in an adverse environment, including stationary and non-stationary noise environments.