Abstract
This study shows our successful attempt to simulate speech mechanism with the aid of an analog computer, thereby devising a speech synthesizer which can speak continuousty in real time. We began with developing an ultra-high speed analog computer which can operate in the speech frequency band width. This computer was used for making a vocal tract simulator, making it possible to synthesize Japanese vowels of fairly good quality. Next, in synthesizing Japanese consonants, the parameters corresponding to conrigurations were controlled dynamically by the use of a newly developed high speed analog multiplier. As a result, control noises were eliminated and smooth control was ensured, thus making it possible to synthesize all of fifty Japanese alphabets. By the way, in order to get continuously synthesized voices, the hybrid system was employed for control. Our new method has been producing a satisfactory result in synthesizing songs as well as speeches.
