Recognition Performance Improvement of Unsupervised Limabeam Algorithm using Post Filtering Technique
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
In distant-talking environments, speech recognition performance degrades significantly due to noise and reverberation. Recent work of Michael L. Selzer shows that in microphone array speech recognition, the word error rate can be significantly reduced by adapting the beamformer weights to generate a sequence of features which maximizes the likelihood of the correct hypothesis. In this approach, called Likelihood Maximizing Beamforming algorithm (Limabeam), one of the method to implement this Limabeam is an UnSupervised Limabeam(USL) that can improve recognition performance in any situation of environment. From our investigation for this USL, we could see that because the performance of optimization depends strongly on the transcription output of the first recognition step, the output become unstable and this may lead lower performance. In order to improve recognition performance of USL, some post-filter techniques can be employed to obtain more correct transcription output of the first step. In this work, as a post-filtering technique for first recognition step of USL, we propose to add a Wiener-Filter combined with Feature Weighted Malahanobis Distance to improve recognition performance. We also suggest an alternative way to implement Limabeam algorithm for Hidden Markov Network (HM-Net) speech recognizer for efficient implementation. Speech recognition experiments performed in real distant-talking environment confirm the efficacy of Limabeam algorithm in HM-Net speech recognition system and also confirm the improved performance by the proposed method.
Abstract
In distant-talking environments, speech recognition performance degrades significantly due to noise and reverberation. Recent work of Michael L. Selzer shows that in microphone array speech recognition, the word error rate can be significantly reduced by adapting the beamformer weights to generate a sequence of features which maximizes the likelihood of the correct hypothesis. In this approach, called Likelihood Maximizing Beamforming algorithm (Limabeam), one of the method to implement this Limabeam is an UnSupervised Limabeam(USL) that can improve recognition performance in any situation of environment. From our investigation for this USL, we could see that because the performance of optimization depends strongly on the transcription output of the first recognition step, the output become unstable and this may lead lower performance. In order to improve recognition performance of USL, some post-filter techniques can be employed to obtain more correct transcription output of the first step. In this work, as a post-filtering technique for first recognition step of USL, we propose to add a Wiener-Filter combined with Feature Weighted Malahanobis Distance to improve recognition performance. We also suggest an alternative way to implement Limabeam algorithm for Hidden Markov Network (HM-Net) speech recognizer for efficient implementation. Speech recognition experiments performed in real distant-talking environment confirm the efficacy of Limabeam algorithm in HM-Net speech recognition system and also confirm the improved performance by the proposed method.
참고문헌 (19)
- M.F. Font, "Multi-microphone signal processing for automatic speech recognition in meeting rooms", 석사, pp.2005
- P.J. Moreno, "Speech recognition in noisy environments", 박사, pp.1996
- F.H. Liu, "Environmental adaptation for robust speech recognition", 박사, pp.1994
- A. Acero, "Acoustical and environmental robustness in automatic speech recognition", 박사, pp.1990
- L.J. Griffiths, "An alternative approach to linearly constrained adaptive beamforming", IEEE Transaction on Antennas and Propagation, AP-30(1), pp.27~34 (1982)
- M. Seltzer, "Microphone Array Processing for Robust Speech Recognition", 박사, pp.2003
- P. Placeway, "The 1996 hub-4 sphinx-3 system", Proceedings on the DARPA Speech Recognition workshop, 1, pp.243~252 (1996)
- S.J. Oh, "A study on statistical language models for large vocabulary continuous speech recognition system", Proceedings on ICSP, 1, pp.113~119 (1999)
- W.H. Press, "Numerical Recipes in C: The Art of Scientific Computin" (1998), Cambridge University Press
- L. Rabiner, "Fundamentals of Speech Recognition" (1993), Prentice Hall
- A.J. Viterbi, "Error bounds for convolutional codes and an asymptotically optimum decoding algorithm", IEEE Transactions on Information Theory, 13(2), pp.260~269 (1967)
- P. Morento, "A Unified Approach for Robust Speech Recognition", Proceedings of Eurospeech, 1, pp.481~485 (1995)
- R. Zenlinski, "A microphone array with adaptive post-filtering for noise reduction in reverberant room", Proceedings on International Conference of Acoustics, Speech, and Signal Processing, 5, pp.2578~2581 (1988)
- I.A. McCowan, "Robust speech recognition using microphone arrays", 박사, pp.2001
- C.H. Knapp, "The generalized correlation method for estimation of time delay", IEEE Transactions on Acoustics, Speech and Signal Processing, 24(4), pp.320~327 (1976)
- D.C. Nguyen, "Performance Improvement of Microphone Array Speech Recognition using Feature Weighted Mahalanobis Distance", The Journal of the Acoustical Society of Korea, 29(1E), pp.45~53 (2010)
- N.D. Cuong, "Performace improvement of speech recognition system using microphone array", Proceedings on IEEE International Conference of Research, Innovation and Vision for the Future, pp.91~95 (2008)
- L. Brayda, "N-Best Parallel Maximum Likelihood Beamformers for Robust Speech Recognition", Proceedings of European Signal Processing Conference (2006)
- K. Bahram, "Improvement in speech recognition using phone-based filter and sum parameter optimuization", IEICE Electronics Express, 6(8), pp.437~442 (2009)