In this paper, a new decision tree-based clustering technique called Phonetic, Dimensional and State Positional Decision Tree (PDS-DT) is proposed. In PDS-DT, phonetic contexts, dimensions and state positions are grouped simultaneously during decision tree construction. PDS-DT provides a complicate distribution sharing structure without any external control parameters. In speaker-independent continuous speech recognition experiments, PDS-DT achieved about 13%-15% error reduction over the phonetic decision tree-based state-tying technique.
Cite as: Zen, H., Tokuda, K., Kitamura, T. (2003) Decision tree-based simultaneous clustering of phonetic contexts, dimensions, and state positions for acoustic modeling. Proc. 8th European Conference on Speech Communication and Technology (Eurospeech 2003), 3189-3192, doi: 10.21437/Eurospeech.2003-797
@inproceedings{zen03_eurospeech,
author={Heiga Zen and Keiichi Tokuda and Tadashi Kitamura},
title={{Decision tree-based simultaneous clustering of phonetic contexts, dimensions, and state positions for acoustic modeling}},
year=2003,
booktitle={Proc. 8th European Conference on Speech Communication and Technology (Eurospeech 2003)},
pages={3189--3192},
doi={10.21437/Eurospeech.2003-797}
}