This paper presents a method of modeling a speakers pronunciation of a given language as a blend of "standard" speech and other non-standard speech varieties (regional dialects and foreign accented pronunciation styles) by way of speaker-dependent modification of a lexicon. In this system, a lexicon of Standard American English (SAE) forms, the "canonical" lexicon, is filtered and transformed via a group of speech variety (SV) dependent rule sets into a speaker specific set of pronunciation variants (and associated probabilities) for use during recognition. The relative importance of these rule sets depends on the speakers pronunciation characteristics and is represented by a Speech Variety Pro- file (SVP) associated with each speaker. A speakers individual SVP is acquired through feedback from an adaptation process. Convergence to a speakers SVP represents adaptation of the lexicon (symbolic adaptation) to those SVspecific forms that speaker is likely to utter.
Cite as: Lee, K.-T., Melnar, L., Talley, J. (2002) Symbolic speaker adaptation for pronunciation modeling. Proc. ITRW on Pronunciation Modeling and Lexicon Adaptation for Spoken Language Technology (PMLA 2002), 24-29
@inproceedings{lee02_pmla,
author={Kyung-Tak Lee and Lynette Melnar and Jim Talley},
title={{Symbolic speaker adaptation for pronunciation modeling}},
year=2002,
booktitle={Proc. ITRW on Pronunciation Modeling and Lexicon Adaptation for Spoken Language Technology (PMLA 2002)},
pages={24--29}
}