Abstract
Recent advances in Hidden Markov Model (HMM) based speaker-independent connected digit recognition have usually tended to make the models more complex. This paper concentrates on improving the training techniques in order to make the most of the available parameters. A new algorithm, Corrective MMI Training is introduced. Use of this algorithm resulted in significant improvements in our recognition rates. We now obtain less than 2% string error rate using semi-continuous HMMs with two models per digit.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
G.R. Doddington, Phonetically Sensitive Discriminants for Improved Speech Recognition, ICASSP-89, paper S10b.11
P.S. Gopalakrishnan, D. Kanevsky, A. Nadas and D. Nahamoo, A Generalization of the Baum Algorithm to Rational Objective Functions, ICASSP-89, paper S12.9
B. Merialdo, Phonetic Recognition using Hidden Markov Models and Maximum Mutual Information Training, ICASSP-88, paper S3.4
L.R. Rabiner, C.H. Lee, B.H. Juang, and J.G. Wilpon, HMM Clustering for Connected Word Recognition, ICASSP-89, paper S8.5
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1992 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Normandin, Y., Cardin, R. (1992). Developments in High-Performance Connected Digit Recognition. In: Laface, P., De Mori, R. (eds) Speech Recognition and Understanding. NATO ASI Series, vol 75. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-76626-8_8
Download citation
DOI: https://doi.org/10.1007/978-3-642-76626-8_8
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-76628-2
Online ISBN: 978-3-642-76626-8
eBook Packages: Springer Book Archive