Text-to-Speech Conversion of Standard Malay

Abstract

This article describes an unrestricted vocabulary Text-to-Speech (TTS) conversion system for the segmental synthesis of Standard Malay (SM) speech. The system uses a modified version of a synthesis method that was previously used to synthesize Arabic. SM bears some resemblance to Arabic in its phonetic structure, and this helped to ease the synthesis problem of SM using previous knowledge gained from working with Arabic. However, SM is different from Arabic in many respects. The method is, therefore, enhanced to suit SM after studying its phonetic properties. All other aspects of the segmental TTS conversion process will be dealt with in this article. These include rules for text preprocessing, letter-to-sound or grapheme-to-phoneme transcription, synthesis units' generation, strategies for the production of reasonably intelligible speech, results of intelligibility tests, and suggestions for future research directions.