The Spectrum Editor System With DSP Speech Synthesizer

In order to analyze the phonemic characteristics of speech sounds, we often need to perform a listening test using special kinds of speech data which are synthesized by a computer. Genarally, it takes a long time to prepare speech data for the listening test because a new program is required when we need new speech data. In this paper, we report on a new speech processing system which provides an easy way to synthesize artificial speech sounds using a personal computer. In this system, the speech sound is first analyzed into pitch, power, and spectrum structure, the three basic factors of sound. Next, the time pattern of these three factors are freely and easily edited using a keyboard and a mouse just as when sentences are edited on a word processor. Last, we synthesize the speech sound from these three edited factors. Because we process the three basic factors of sound, we can synthesize any kind of speech sound on this system