We report on an automatic method for discovering an appropriate model topology for each context-dependent phoneme, allowing for such phenomena as reduced pronunciations and substituted phonemes. The method leads to a reduction in the word error rate on both the Wall Street Journal and Broadcast News databases.
Cite as: Eide, E. (1999) Automatic modeling of pronunciation variations. Proc. 6th European Conference on Speech Communication and Technology (Eurospeech 1999), 451-454, doi: 10.21437/Eurospeech.1999-116
@inproceedings{eide99_eurospeech,
author={Ellen Eide},
title={{Automatic modeling of pronunciation variations}},
year=1999,
booktitle={Proc. 6th European Conference on Speech Communication and Technology (Eurospeech 1999)},
pages={451--454},
doi={10.21437/Eurospeech.1999-116}
}