Noise-Robust Speaker Recognition Using Subband Likelihoods and Reliable-Feature Selection

ETRI Journal

关键词: adaptive noise model;    reliable feature selection;    subband likelihood;    mel-frequency cepstral coefficient;    feature recombination;    universal background model;    Gaussian mixture model;    Speaker recognition;   

Others  :  1185720
DOI  :  10.4218/etrij.08.0107.0108

【 摘 要 】

We consider the feature recombination technique in a multiband approach to speaker identification and verification. To overcome the ineffectiveness of conventional feature recombination in broadband noisy environments, we propose a new subband feature recombination which uses subband likelihoods and a subband reliable-feature selection technique with an adaptive noise model. In the decision step of speaker recognition, a few very low unreliable feature likelihood scores can cause a speaker recognition system to make an incorrect decision. To overcome this problem, reliable-feature selection adjusts the likelihood scores of an unreliable feature by comparison with those of an adaptive noise model, which is estimated by the maximum a posteriori adaptation technique using noise features directly obtained from noisy test speech. To evaluate the effectiveness of the proposed methods in noisy environments, we use the TIMIT database and the NTIMIT database, which is the corresponding telephone version of TIMIT database. The proposed subband feature recombination with subband reliable-feature selection achieves better performance than the conventional feature recombination system with reliable-feature selection.

   

【 参考文献 】

• [1]D. Reynold and R.C. Rose, "Robust Text Independent Speaker Identification Using Gaussian Mixture Speaker Models," Proc. IEEE Tran. Speech and Audio Processing, vol. 3, Jan. 1995, pp. 72-83.
• [2]D. Reynolds, T. Quatieri, and R. Dunn, "Speaker Verification Using Adapted Gaussian Mixture Models," Digital Signal Processing, vol. 10, 2000, pp. 19-41.
• [3]A. Drygajlo and M. El-Maliki, "Speaker Verification in Noisy Environments with Combined Spectral Subtraction and Missing Feature Theory," Proc. ICASSP, vol. 2, 1998, pp. 121-124.
• [4]C. Barras and J. Gauvain, "Feature and Score Normalization for Speaker Verification of Cellular Data," Proc. ICASSP, 2003, pp. 49-52.
• [5]K. Yiu, M. Mak, and S. Kung, "Environment Adaptation for Robust Speaker Verification," Proc. EUROSPEECH, 2003, pp. 2973-2976.
• [6]H.J. Qing, Z. Lei, and W. Chengfa, "An Environment Adaptation Method for Robust Speech Recognition," Proc. ICSP, 2000, pp. 726-729.
• [7]D. Ramos-Castro, J. Fierrez-Aquilar, J. Gonzalez-Rodriquez, and J. Ortega-Garcia, "Speaker Verification Using Speaker- and Test-Dependent Fast Score Normalization," Pattern Recognition Letters, vol. 28, 2007, pp. 90-98.
• [8]R. Auckenthaler, M. Carey, and H. Lloyd-Thomas, "Score Normalization for Text-Independent Speaker Verification Systems," Digital Signal Processing, vol. 10, 2000, pp. 42-54.
• [9]S. Okawa, E. Bocchieri, and A. Potamianos, "Multiband Speech Recognition in Noise Environments," Proc. ICASSP, 1998, pp. 641-644.
• [10]H. Hermansky, S. Tibrewala, and M. Pavel, "Toward ASR on Partially Corrupted Speech," Proc. ICSLP, 1996.
• [11]S. Tibrewala and H. Hermansky, "Subband Based Recognition of Noisy Speech," Proc. ICASSP, 1997, pp. 1255-1258.
• [12]W. Chen, C. Hsieh, and E. Lai, "Multiband Approach to Robust Text-Independent Speaker Identification," Computational Linguistics and Chinese Language Processing, vol. 9, no. 2, 2004, pp. 63-76.
• [13]B. Mak, "A Mathematical Relationship Between Full-Band and Multiband Mel-Frequency Cepstral Coefficients," IEEE Signal Processing Letters, vol. 9, no. 8, 2002, pp. 241-244.
• [14]D. Reynold, "Large Population Speaker Identification Using Clean and Telephone Speech," IEEE Signal Processing Letters, vol. 2, no. 3, 1995, pp. 46-48.
• [15]D. Pearce and H. Hirsch, "The Aurora Experimental Framework for the Performance Evaluation of Speech Recognition Systems under Noise Conditions," Proc. ICSLP, vol. 4, 2000, pp. 29-32.