For the Realization of Mechanical Speech Synthesizer. Realization of Three-Dimensional Shapes of Articulatory Organs.

In this paper, we suggest the means of speech synthesis using Mechanical Speech Synthesizer which seems to be suitable for humanoid robot. The present speech synthesis by software systems or electrical circuits do not sufficiently take the mechanism of human speech production into consideration. Therefore, in this research, we make the system of speech synthesis using the model which has human-like regions related with speech production. We synthesize speech by controlling the model's regions as a man controls. As this system will be able to reproduce the physical phenomenon during speech production, this means of speech synthesis is expected to be able to synthesize natural speech as a man produces.For the first step to our research, we made a clay model of vocal tract, and estimated the three-dimensional shapes of its tongue and lips for synthesizing Japanese steady vowels.