Abstract
Hybrid methods which combine hidden Markov models (HMMs) and connectionist techniques take advantage of what are believed to be the strong points of each of the two approaches: the powerful discrimination-based learning of connectionist networks and the time-alignment capability of HMMs. Connectionist Viterbi Training (CVT) is a simple variation of Viterbi training which uses a back-propagation network to represent the output distributions associated with the transitions in the HMM. The work reported here represents the culmination of three years of investigation of various means by which HMMs and neural networks (NNs) can be combined for continuous speech recognition. This paper describes the CVT procedure, discusses the factors most important to its design and reports its recognition performance. Several changes made to the system over the past year are reported here, including: (1) the change from recurrent to non-recurrent NNs, (2) the change from Sphinx-style phone-based HMMs to word-based HMMS, (3) the addition of a corrective training procedure, and (3) the addition of an alternate model for every word. The CVT system, incorporating these changes, achieves 99.1% word accuracy and 98.0% string accuracy on the TI/NBS Connected Digits task (“TI Digits”).
Preview
Unable to display preview. Download preview PDF.
References
Brown, P., The Acoustic Modeling Problem in Automatic Speech Recognition, Ph.D. Thesis, Carnegie Mellon University, May, 1987.
Huang, W., Lippmann, R. “HMM Speech Recognition with Neural Net Discrimination,” Proc. Neural Information Processing Systems (NIPS) Conference, November, 1989.
Bourlard, H. and Morgan, N. Merging Multilayer Perceptrons and Hidden Markov Models: Some Experiments in Continuous Speech Recognition, Tech. Report TR-89-033, July, 1989, International Computer Science Institute, Berkeley, CA.
Elman, J.L. Finding Structure in Time, Tech. report, Center for Research in Language, University of California, San Diego, April, 1988.
Bakis, R. “Continuous Speech Recognition via Centisecond Acoustic States,” Proc. 91st Meeting Acoustical Soc. of America. April, 1976.
Picone, J. “On Modeling Duration in Context in Speech Recognition,” Proc. ICASSP, April, 1989.
Lee, K.F. Large-Vocabulary Speaker-Independent Continuous Speech Recognition: The SPHINX System, Ph.D. Thesis, Carnegie Mellon University. 1988.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1991 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Franzini, M., Waibel, A., Lee, KF. (1991). Continuous speech recognition with the connectionist viterbi training procedure: A summary of recent work. In: Prieto, A. (eds) Artificial Neural Networks. IWANN 1991. Lecture Notes in Computer Science, vol 540. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0035914
Download citation
DOI: https://doi.org/10.1007/BFb0035914
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-54537-8
Online ISBN: 978-3-540-38460-1
eBook Packages: Springer Book Archive