Rapid Heteronym Disambiguation for Text-to-Speech System

This paper is about the text analysis component for heteronym words for Malay Text-to-Speech (TTS) system. Heteronym is a homograph words which the meaning and the pronunciation differ according to the semantic of the sentence. There is only 12 unique heteronym in Malay. However, these words can be expanded depending on the morphology of the sentence. Malay language is also very rich with prefix and suffixes. Thus, one word can be expanded either to two unique words or more. For the list of 12 unique pair of words, we can expand the list to 98 unique words. Thus, there are greater variation possibility of enunciating a heteronym words. This research showed the importance of a study on heteronym words specifically in TTS case as well as what would be the best approach of handling heteronym ambiguity when only the sampled data is available but not tagged data of a particular target language.