The effect of syllable and word stress on the quality of Indonesian HMM-based speech synthesis system

A speech synthesis system enables the generation of speech waveform from textual data automatically. In this paper, we report our study on speech synthesis system for Indonesian language. The speech synthesis system uses HMM which have proven to be an efficient parametric model to synthesize speech waveform. In this study, we analyze the effect of syllable and word stress on the quality of synthesized speech. Result shows that the use of syllable and word stress can improve the quality of the synthesized speech with MOS level for about 67.3% and word accuracy level for about 0.01%.