As grapheme-to-phoneme methods proliferate, their careful evaluation becomes increasingly important. This paper explores a variety of metrics to compare the automatic pronunciation methods of three freely-available grapheme-to-phoneme packages on a large dictionary. Two metrics, presented here for the first time, rely upon a novel weighted phonemic substitution matrix constructed from substitution frequencies in a collection of trusted alternate pronunciations. These new metrics are sensitive to the degree of mutability among phonemes. An alignment tool uses this matrix to compare phoneme substitutions between pairs of pronunciations.
Cite as: Hixon, B., Schneider, E., Epstein, S.L. (2011) Phonemic similarity metrics to compare pronunciation methods. Proc. Interspeech 2011, 825-828, doi: 10.21437/Interspeech.2011-305
@inproceedings{hixon11_interspeech, author={Ben Hixon and Eric Schneider and Susan L. Epstein}, title={{Phonemic similarity metrics to compare pronunciation methods}}, year=2011, booktitle={Proc. Interspeech 2011}, pages={825--828}, doi={10.21437/Interspeech.2011-305} }