Pronunciation modelling of Penang Hokkien dialect for text-to-speech system

• Main Author: Lim, Kang Jie
• Format: Final Year Project / Dissertation / Thesis
• Published: 2022
• Subjects: T Technology (General)
• Online Access: http://eprints.utar.edu.my/4728/1/fyp_IA_2022_LKJ.pdf; http://eprints.utar.edu.my/4728/

This is academic research about pronunciation modelling of Penang Hokkien for Text-to-Speech System which is under field of study, Speech Synthesis. It is widely known that there are majority of unwritten languages are gradually forgotten by younger generations due to domination of written languages in education and the most significant factor is lack of documentation of the languages. Hence, these hindrances prevent or increase the effort of revitalization on those unwritten languages by implementing current technologies. Penang Hokkien Language, a likely unwritten language spoke in Northern of Southern Peninsular Malaysia is selected as case study of this research where its linguistic resources are partially documented. In order to develop an TTS System for Penang Hokkien, this research project is the first steps to familiarize with this high complexity language. Since this project is part of the effort in revitalizing Penang Hokkien Language, Traditional Chinese Character is opted as standard of writing system and Penang Hokkien Spelling System which created by Hokkien Association of Penang is selected as standard of pronunciation orthography. Listing of phonemes with categorizing them into initials and finals are taken as Penang Hokkien is a tonal language. Moreover, nine tones are marked with the use of diacritics based on Penang Hokkien Spelling System according to the tone marking rules. Tone sandhi rules are also created in orthography standardization phase. The contributions of this project are (1) finding the possible combinations of initials and finals and tones, (2) collect possible graphemes, (3) map graphemes with morphemes, (4) design database to store the processed data and (5) standardizing the tones and tone sandhi rules.