Abstract

This study is intended to develop a noise robust distributed speech recognizer for real-world applications by employing Cepstral Mean Normalization (CMN) for robust feature extraction. The main focus of the work is to cope with different noisy environments. To realize this objective, Mel-LP based speech analysis has been used in speech coding on the linear frequency scale by applying a first-order all-pass filter instead of a unit delay. Mismatch between training and test phases is reduced through robust feature extraction by applying CMN on Mel-LP cepstral coefficients as an effort to reduce additive noise and channel distortion. The performance of the proposed system has been evaluated on test set A of Aurora-2 database which is a subset of TIDigits database contaminated by additive noises and channel effects. The experiment is conducted on four different noisy environments and the baseline performance, that is, for Mel-LPC the average word accuracy has found to be 59. 05%. By applying the CMN on Mel-LP cepstral coefficients, the performance has been improved to 68. 02%. It is found that CMN performs significantly better for different noisy environments.

References

• Bateman, D. C. , et al. , 1992. Spectral contrast normalization and other techniques for speech recognition in noise. Proc. of ICASSP '92, I: 241-244.
• Vaseghi, S. V. and B. P. Milner, 1993. Noise-adaptive hidden Markov models based on Wiener filters. Proc. of Eurospeech '93, II: 1023-1026.
• Islam, M. B. , K. Yamamoto, H. Matsumoto, 2007. Mel-Wiener filter for Mel-LPC based speech recognition. IEICE Transactions on Information and Systems, E90-D (6): 935-942.
• Boll, S. F. , 1979. Suppression of acoustic noise in speech using spectral subtraction. IEEE Trans. Acoust. , Speech and Signal Processing, 27(2): 113-120.
• Lim, J. S. and A. V. Oppenheim, 1979. Enhancement and bandwidth compression of noisy speech. Proc. of the IEEE, 67(2): 1586-1604.
• Lockwood, P. and J. Boudy, 1992. Experiments with a nonlinear spectral subtractor (nss), hidden Markov models and the projection or robust speech recognition in cars. Speech Commun. , 11(2-3): 215-228.
• Agarwal, A. and Y. M. Cheng, 1999. Two-stage Mel-warped Wiener filter for robust speech recognition. Proc. of ASRU '99: 67-70.
• Zhu, Q. and A. Alwan, 2002. The effect of additive noise on speech amplitude spectra: A Quantitative analysis. IEEE Signal Processing Letters, 9(9): 275-277.
• Atal, B. , 1974. Effectiveness of linear prediction characteristics of the speech wave for automatic speaker identification and verification. J. Acoust. Soc. Am. 55(6): 1304-1312.
• Furui, S. , 1981. Cepstral analysis technique for automatic speaker verification. IEEE Trans. Acoust. , Speech and Signal Processing, ASSP-29: 254-272.
• Mokbel, C. , et al. , 1984. Compensation of telephone line effects for robust speech recognition. Proc. of ICSLP '94: 987-990.
• Gales, M. J. F. and S. J. Young, 1993a. HMM recognition in noise using parallel model combination. Proc. of Eurospeech '93, II: 837-840.
• Gales, M. J. F. and S. J. Young, 1993b. Cepstral parameter compensation for HMM recognition in noise. Speech Communication, 12(3): 231-239.
• Varga, A. P. and R. K. Moore, 1990. Hidden Markov model decomposition of speech and noise. Proc. of ICASSP '90, 2: 845-848.
• Davis, S. and P. Mermelstein, 1980. Comparison of parametric representations for monosyllabic word recognition in continuously spoken sentences. IEEE Trans. on Acoustics, Speech, and Signal Processing, ASSP-28(4): 357-366.
• Hermansky, H. , 1987. Perceptual linear predictive (PLP) analysis of speech. J. Acoust. Soc. Am. , 87(4): 17-29.
• Virag, N. , 1995. Speech enhancement based on masking properties of the auditory system. Proc. ICASSP '95: 796-799.
• Itakura, F. and S. Saito, 1968. Analysis synthesis telephony based upon the maximum likelihood method. Proc. of 6th International Congress on Acoustics, Tokyo: C-5-5, C17-20.
• Atal, B. and M. Schroeder, 1968. Predictive coding of speech signals. Proc. of 6th International Congress on Acoustics, Tokyo: 21-28.
• Makhoul, J. and L. Cosell, 1976. LPCW: An LPC vocoder with linear predictive warping. Proc. of ICASSP '76: 446-469.
• Itahashi, S. and S. Yokoyama, 1987. A formant extraction method utilizing mel scale and equal loudness contour. Speech Transmission Lab. -Quarterly Progress and Status Report (Stockholm) (4): 17-29.
• Rahim, M. G. and B. H. Juang, 1996. Signal bias removal by maximum likelihood estimation for robust telephone speech recognition. IEEE Trans. on Speech and Audio Processing, 4(1): 19-30.
• Strube, H. W. , 1980. Linear prediction on a warped frequency scale. J. Acoust. Soc. Am. , 68(4): 1071-1076.
• Oppenheim, A. V. and D. H. Johnson, 1972. Discrete representation of signals. IEEE Proc. , 60(6): 681-691.
• Matsumoto, H. , et al. , 1998. An efficient Mel-LPC analysis method for speech recognition", Proc. ICSLP '98: 1051-1054.
• Zwicker, E. and E. Terhardt, 1980. Analytical expressions for critical band rate and critical bandwidth as a function. J. Acoust. Soc. Am. , 68: 1523-1525.
• Lindsay, P. H. and D. A. Norman, 1977. Human information processing: An introduction to psychology. 2nd Edn. , Academic Press.
• Nakagawa, S. , et al. , ed. , 2005. Spoken language systems. Ohmsha, Ltd. , Japan, ch. 7.
• Markel, J. and A. Gray, 1976. Linear prediction of speech. Springer-Verlag.
• Acero, A. and R. Stern, 1990. Environmental robustness in automatic speech recognition. Proc. of ICASSP '90: 849-852.
• Hirsch, H. G. and D. Pearce, 2000. The AURORA experimental framework for the performance evaluation of speech recognition systems under noisy conditions. Proc. ISCA ITRW ASR 2000: 181:188.
• Leonard, R. G. , 1984. A database for speaker independent digit recognition. ICASSP84, 3: 42. 11.
• ITU recommendation G. 712, 1996. Transmission performance characteristics of pulse code modulation channels.