Robust speech recognition features based on temporal trajectory filtering of frequency band spectrum

This paper presents the use of a variety of lters in the temporal trajectories of frequency band spectrum to extract speech recognition features for environmental robustness. Three kind of lters for emphasizing the statistically important parts of speech are proposed. First, a bank of RASTA-like band-pass lters to t the statistical peaks of modulation frequency band spectrum of speech are used. Secondly, a three-channel octave bandlter band with a smoothed rectangular window spline is applied. Thirdly, a datadriven lter is developed. Experimental results show that signi cant improvements for speech recognition using the proposed feature extraction approach under noisy environments can be achieved.