Grapheme to Phoneme Transcription for Pali-Thai

Abstract

This paper presents a method for Pali phoneme transcription using a hybrid of dictionary and linguistic rules. In the study, we constructed a Pali-Thai dictionary, which gathered Pali words from the dictionary book of Phramahapriroch. The dictionary consists of almost 14,323 Pali words. The dictionary structure contains Pali words, meanings, word type, corresponding Thai words, romanize words, as well as phoneme. The input of the developed system is Pali language in paragraph level. The input sentence was segmented into sequence of words using the longest matching algorithm in order to look for Pali word and phoneme from the dictionary. Words that do not appear in the dictionary are transcribed using the Pali grammar and linguistic rules. In order to evaluate the efficiency of the proposed system, the developed system was used to transcribe Pali documents. The testing documents are 99 prayers, 181 proverbs and Tipitaka Volume 1 of siamrath. The transcription result shows that the accuracy
the transcription was 98.75%.