Feature Compensation Combining SNR-Dependent Feature Reconstruction and Class Histogram Equalization

ETRI Journal

关键词: robust speech recognition;    feature reconstruction;    Class histogram equalization;   

Others  :  1185651
DOI  :  10.4218/etrij.08.0208.0147

【 摘 要 】

In this letter, we propose a new histogram equalization technique for feature compensation in speech recognition under noisy environments. The proposed approach combines a signal-to-noise-ratio–dependent feature reconstruction method and the class histogram equalization technique to effectively reduce the acoustic mismatch present in noisy speech features. Experimental results from the Aurora 2 task confirm the superiority of the proposed approach for acoustic feature compensation.

   

【 参考文献 】

• [1]X. Huang, A. Acero, and H.-W. Hon, Spoken Language Processing: A Guide to Theory, Algorithm, and System Development, Upper Saddle River, NJ: Prentice Hall, 2001.
• [2]J.C. Segura et al., "Cepstral Domain Segmental Nonlinear Feature Transformations for Robust Speech Recognition," IEEE Signal Processing Letters, vol. 11, no. 5, 2004, pp. 517-520.
• [3]A. de la Torre et al., "Histogram Equalization of Speech Representation for Robust Speech Recognition," IEEE Trans. Speech, Audio Processing, vol. 13, no. 3, 2005, pp. 355-366.
• [4]Y. Suh and H. Kim, "Class-Based Histogram Equalization for Robust Speech Recognition," ETRI J., vol. 28, no. 4, Aug. 2006, pp. 502-505.
• [5]Y. Suh, M. Ji, and H. Kim, "Probabilistic Class Histogram Equalization for Robust Speech Recognition," IEEE Signal Processing Letters, vol. 14, no. 4, Apr. 2007, pp. 287-290.
• [6]J.C. Segura et al., "Feature Extraction Combining Spectral Noise Reduction and Cepstral Histogram Equalization for Robust ASR," Proc. ICSLP, Sept. 2002, pp. 225-228.