KORPUS PERTUTURAN SINTAKSIS-PROSODI BAHASA MELAYU

This paper presents a construction of a Malay syntax-prosody speech corpus that was intentionally built for a Malay concatenative speech synthesizer system. The Malay syntax-prosody speech corpus was represented by the syntax-prosody tree structures representation, an adaptation of a syntactic tree structure representation called Structured-String Tree Correspondence (SSTC). In order to build the Malay speech corpus in syntaxprosody representation, existing sentences of SSTC were used for recording. Prosodic features, which were obtained from the recording files, were annotated to the SSTC tree structure and at the same time, sound files were linked to the SSTC trees. The speech corpus contains 422 sentences, 1720 phrase and 6978 word units.