| 摘要 |
PROBLEM TO BE SOLVED: To make sounds audible smooth and natural by using the same channel in the case of voiced sounds with each other and unvoiced sound with each other and lowering a level and switching the channel by an envelope generator(EG) to tactfully connect the breaks of phonemes and phonemes in other cases. SOLUTION: In the case of the transition from the voiced sounds to the voiced sounds, from the unvoiced sounds to the unvoiced sounds and from the unvoiced sounds to the voiced sounds, parameter values are formed by executing interpolation processing from the central frequency of Formant of the preceding phoneme and the value of a Formant level to the succeeding phoneme and are sent from a CPU to a Formant synthesizing sound source 110. On the other hand, in the case of the transition from the voiced sounds to the unvoiced sounds, the case the Formant level is rapidly lowered is recognized by the relation of the front and rear phonemes and the Formant level of the preceding phoneme is lowered by using the envelope waveforms outputted at every sampling period from the EGs 114, 214 disposed in the inside. The succeeding phonemes are then produced by switching the sound source channels. |
