Robust MVDR-based feature extraction for speech recognition

This paper presents a novel noise robust feature extraction method for speech recognition. It is based on making the Minimum Variance Distortionless Response (MVDR) power spectrum estimation method robust against noise. This robustness is obtained by modifying the distortionless constraint of the MVDR spectral estimation method via weighting the subband power spectrum values based on the sub-band signal to noise ratios. The above method, when evaluated on Aurora 2 task, outperformed both the MFCC features as the baseline and the MVDR-based features in different noisy conditions.

[1]  David Pearce,et al.  The aurora experimental framework for the performance evaluation of speech recognition systems under noisy conditions , 2000, INTERSPEECH.

[2]  Kuang Jingming,et al.  NOISE SUPPRESSION BASED ON TEAGER ENERGY OPERATOR FOR IMPROVING THE ROBUSTNESS OF ASR FRONT-END , 2003 .

[3]  W. M. Carey,et al.  Digital spectral analysis: with applications , 1986 .

[4]  S.M. Ahadi,et al.  Weighting of Mel Sub-bands Based on SNR/Entropy for Robust ASR , 2008, 2008 IEEE International Symposium on Signal Processing and Information Technology.

[5]  S. Seyedin,et al.  Feature extraction based on DCT and MVDR spectral estimation for robust speech recognition , 2008, 2008 9th International Conference on Signal Processing.

[6]  Shu Hung Leung,et al.  SNR-dependent non-uniform spectral compression for noisy speech recognition , 2004, 2004 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[7]  H Hermansky,et al.  Perceptual linear predictive (PLP) analysis of speech. , 1990, The Journal of the Acoustical Society of America.

[8]  Bhaskar D. Rao,et al.  Robust Feature Extraction for Continuous Speech Recognition Using the MVDR Spectrum Estimation Method , 2007, IEEE Transactions on Audio, Speech, and Language Processing.