2026-06-25T03:18:04Z https://meral.edu.mm/oai

oai:meral.edu.mm:recid/4882 2022-03-24T23:13:38Z 1582963302567:1597824273898 user-ucsy

Syllable Pronunciation Features for Myanmar Grapheme to Phoneme Conversion Thu, Ye Kyaw Pa, Win Pa Finch, Andrew Hlaing, Aye Mya Naing, Hay Mar Soe Sumita, Eiichiro Hori, Chiori Grapheme-to-Phoneme (G2P) conversion is anecessary step for speech synthesis and speechrecognition. This paper addresses the problem ofgrapheme to phoneme conversion for the Myanmarlanguage. In our method, we propose four simpleMyanmar syllable pronunciation patterns as featuresthat can be used to augment the models in a ConditionalRandom Field (CRF) approach to G2P conversion. Ourresults show that our additional features are able toimprove a strong baseline model that does not includethem. We found that combination of all four featuresgave rise to the highest performancefor Myanmarlanguage G2P conversion. 2015-02-05 http://hdl.handle.net/20.500.12678/0000004882 https://meral.edu.mm/records/4882