Development of a Japanese and English Speech Synthesis System Based on HMM Using Voice Conversion for the People with Speech Communication Disorder

In this paper, we describe our work on developing a bilingual communication aid for a Japanese amyotrophic lateral sclerosis (ALS) patient. For this study, HTS toolkit was used to develop the system. First developed was a Japanese speech synthesis and two methods were attempted. The first used an acoustic model built from the recordings of the patient’s read speech. The second used an acoustic model built from a voice converted to the patient’s voice. The result of the perceptual experiment showed that the voice synthesized with the latter was perceived to have a closer voice quality to the patient’s natural speech. An acoustic model for English speech synthesis system was built and GUI works on windows was developed for the patient to use.