Treehouse Talk 12/4/2008
Hierarchical Exception-Marking for Automated Transcription of Thai
I present a system for automatic transcription of Thai script into any of six phonemic representations (phonemic Thai script, plus five different Romanization schemes), one phonetic representation (IPA), and one transliteration system (ISO 11940). Beginning from a simple finite state transducer, the system has evolved over eight years to incorporate several layers of provision for manual notation of exception and idiosyncrasy in the spoken Thai language. This talk will review relevant issues in phonemic transcription and Thai orthography, and present a brief history of the project before proceeding to a discussion of the implementation and the exception-notation hierarchy. An online web demonstration will follow. Evaluation metrics versus Thai Royal Institute gold standard will be presented. Finally, future directions for this work will be discussed.
-- Main.gslayden - 15 Nov 2008
Topic revision: r2 - 2008-12-05 - 02:35:33 - gslayden