For large vocabulary recognition system, as well as for flexible vocabulary applications using hidden Markov model(HMM), parameter smoothing and tying have been used to increase the reliability of models. This paper describes bottom-up and top-down clustering techniques for state level tying. This paper also describes a method of applying parameter smoothing to the clustered states and covariance matrix of semicontinuous hidden Markov model(SCHMM). We applied co-occurrence smoothing method(CSM) for senone smoothing. We present a new parameter smoothing method and apply it to the distribution of discrete hidden Markov model(DHMM) in the training procedure. A new model composition method for unseen triphone modeling in bottom-up clustering is also proposed and compared with traditional context-independent model backing-off method.
Cite as: Choi, J.-S., Lee, J.-S., Lee, H.-Y. (1998) Smoothing and tying for Korean flexible vocabulary isolated word recognition. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 0623, doi: 10.21437/ICSLP.1998-286
@inproceedings{choi98_icslp,
author={Jae-Seung Choi and Jong-Seok Lee and Hee-Youn Lee},
title={{Smoothing and tying for Korean flexible vocabulary isolated word recognition}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 0623},
doi={10.21437/ICSLP.1998-286}
}