Improvement of naturalness for an HMM-based Vietnamese speech synthesis using the prosodic information

Natural-sounding synthesized speech is goal of HMM-based Text-to-Speech systems. Besides using context dependent tri-phone units from a large corpus speech database, many prosody features have been used in full-context labels to improve naturalness of HMM-based Vietnamese synthesizer. In the prosodic specification, tone, part-of-speech (POS) and intonation information are considered not as important as positional information. Context-dependent information includes phoneme sequence as well as prosodic information because the naturalness of synthetic speech highly depends on the prosody such as pause, tone, intonation pattern, and segmental duration. In this paper, we propose decision tree questions that use context-dependent tones and investigate the impact of POS and intonation tagging on the naturalness of HMM-based voice. Experimental results show that our proposed method can improve naturalness of a HMM-based Vietnamese TTS through objective evaluation and MOS test.