Paper: Inducing Probabilistic Syllable Classes Using Multivariate Clustering

ACL ID P00-1029
Title Inducing Probabilistic Syllable Classes Using Multivariate Clustering
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2000
Authors

An approach to automatic detection of syllable structure is presented. We demonstrate a novel application of EM-based clustering to multivariate data, exemplied by the induction of 3- and 5-dimensional probabilis- tic syllable classes. The qualitative evaluation shows that the method yields phonologically meaningful syl- lable classes. We then propose a novel approach to grapheme-to-pho- neme conversion and show that syl- lable structure represents valuable information for pronunciation sys- tems.