Pengembangan Aplikasi Text to Speech untuk Bahasa Bali

Abstract

Text to Speech as an software that can
convert written text into spoken word will be very
useful for people who have weak eyesight and people
who want to learn a spesific language, but exist Text
to Speech software that has been developed by other
researchers can not be applied for the Balinese
language, where existance of Balinese language Text
to Speech software is important as a cultural
preservation efforts in the midst of cultural extinction
issue in Indonesia. This research aimed to (1) design a
Balinese language Text to Speech software; (2)
implement the design of Balinese language Text to
Speech software. The research method used is
development. The process of text-to-speech
conversion, designed and implemented with two stage
process: (1) the process of text-to-phonemes
conversion; (2) the process of phonemes-to-speech
conversion that uses diphone concatenation method by
utilizing MBROLA and diphone database ID1
generated by eSpeak. The results of this research is the
Balinese language Text to Speech software that can
convert written text in latin letters of Balinese
language to spoken words and also save the voices to
output files such as audio file with *.wav extension
and MBROLA phoneme data file with *.pho
extension. The software is named "Bali Text to
Speech" that abbreviated as "BaliTTS" and
implemented with the Java programming language.
Experimental results show that the BaliTTS software
can perform its function in terms of both functional
and conceptual/ structural. BaliTTS software is
expected to facilitate a person to learn the
characteristics of speech in Balinese language and
help people with weak eyesight, such as blind people to
read written text in Balinese language rules.