Dong Yu, Li Deng, Yifan Gong, and Alex Acero
Recently we proposed a cubic-spline-based variableparameter hidden Markov model (CS-VPHMM) whose mean and variance parameters vary according to some cubic spline functions of additional environment-dependent parameters. We have shown good properties of the CS-VPHMM and demonstrated on the Aurora-3 corpus that MCE-trained CSVPHMM greatly outperforms the MCE-trained conventional HMM at the cost of increased total number of model parameters. In this paper, we propose to share spline functions across different Gaussian mixture components to reduce the total number of model parameters and develop a clustering algorithm to do so. We demonstrate the effectiveness of our parameter clustering and sharing algorithm for the CSVPHMM on Aurora-3 corpus and show that proper parameter sharing can reduce the number of parameters from 4 times of that used in the conventional HMM to 1.13 times and still get 18% relative WER reduction over the MCE trained conventional HMM under the well-matched condition. Effective parameter sharing makes the CS-VPHMM an attractive model for noise robustness.
Index Terms: speech recognition, variable-parameter hidden Markov model, cubic spline, parameter sharing, clustering
In Proc. of the Interspeech
Publisher International Speech Communication Association
© 2007 ISCA. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the ISCA and/or the author.