Development of a rule-based speech synthesis system for the Japanese language using a MELP vocoder

A rule-based speech synthesis system for the Japanese language, which employs a MELP (Mixed Excitation Linear Prediction) vocoder as its speech synthesizer, was implemented. This paper especially describes some speech synthesis techniques utilized in our system. Since the MELP vocoder developed in this study could effectively gain the naturalness of voiced consonants as well as purely voiced speech, the implemented system could succeed in enhancing the voice quality of synthesized speech more than a system that employed a conventional normal LPC (Linear Predictive Coding) vocoder.