Improving Bahasa Melayu Text-To-Speech System Using Unit-Selection Method : UTeM Health Centre Case

Sri Winarsih, Nurul Anisa (2017) Improving Bahasa Melayu Text-To-Speech System Using Unit-Selection Method : UTeM Health Centre Case. Masters thesis, Universiti Teknikal Malaysia Melaka.

[img] Text
Improving Bahasa Melayu Text-To-Speech System Using Unit-Selection Method Utem Health Centre Case - Nurul Anisa Sri Winarsih - 24 Pages.pdf - Submitted Version

Download (409kB)

Abstract

The main objective of UTeM health center is to provide the best medical facilities for all the community on campus. Nowadays UTeM Health Center already implements a doctor-patient consultation system. In that consultation system has a calling system using Text-to-Speech (TTS) to call patient name to go to doctor's room but lack of natural sounding speech. That is way the research wants to make improvement for calling system UTeM Health Center with a TTS using Unit-Selection Synthesis (USS) method with Malaysian native speaker for the corpus database. The corpus consists of 200 famous Malaysian names and 1500 sentences. Those corpus segmented into word, syllable, and diphone based. TTS system consists of Natural Language Processing (NLP) and Digital Signal Processing (DSP). NLP is a module to produce a phonetic transcription of the read text. In this study, NLP process consists of text normalization and phonetizer. Then followed by DSP process using Unit-Selection Synthesis (USS) to convert phonetic transcriptions from NLP module into speech synthesis. This TTS has been tested by Comparison Category Rating (CCR) from listening test. The first test is to compare the naturalness of the current calling system and the proposed calling system. The proposed system can solved the problem of naturalness because its result is higher than current system. The second testing is to test the improvement of TTS method for Bahasa Melayu. The proposed system give improvement because the total average result score is 2.55 over 3.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Speech synthesis, Automatic speech recognition, Speech processing systems
Subjects: T Technology > T Technology (General)
T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions: Library > Tesis > FTMK
Depositing User: Nor Aini Md. Jali
Date Deposited: 25 Apr 2018 09:20
Last Modified: 25 Apr 2018 09:20
URI: http://eprints.utem.edu.my/id/eprint/20739
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item