Improved Synthesis of Standard Malay

Y.A. El-Imam (UAE) and Z.M. Don (Malaysia)

Keywords

Speech synthesis, TexttoSpeech, Standard Malay, Phonology and Phonetics.

Abstract

A previous method that uses basic sound clusters, or basic synthesis units, to synthesize standard Malay (SM) is reinvestigated in relation to phonetic variations of SM. It is found that the method covers most phonetic variations of SM. However, there is need to enhance the quality of the synthesized speech by considering the contextual phonetic variations of the synthesis clusters. The basic and the phonetic variants of the clusters are derived from the Malay syllables. An enhanced speech database of the basic sound clusters and their phonetic variations is created and tested to control its segmental quality. A computer-based TTS system is developed using the enhanced method. Speech is synthesized by waveform concatenation. The intelligibility of synthesized speech is assessed by standard intelligibility tests that are adapted to suit the phonetic characteristics of SM. The intelligibility tests show a speech quality improvement over the previous method.

Important Links:



Go Back