| 摘要 |
PURPOSE:To obtain a natural synthesized sound with high quality by preserving a sound source waveform in a driving sound source memory setting length including the multiple pitch cycles of phoneme, etc., as unit, and synthesizing the sound source waveform in unit of the length of the multiple pitch cycles when performing synthesis. CONSTITUTION:A first low-pass filter means 3 which low-pass filters the sound source waveform of the driving sound source memory 1 where a sound signal is preserved in unit of the pitch cycles of the sound source waveform obtained by performing reverse filtering, etc., with a prescribed cut-off frequency, and a sound source expansion/compression means 4 which performs the expansion/ compression of a time waveform obtained by the means 3 with a prescribed ratio are provided. Also, a second low-pass filter means 5 which low-pass filters an obtained time waveform with the prescribed ratio is provided, and it is called from the driving sound source memory 1 setting the multiple pitch cycles as the unit. Thereby, it is possible to express the sensible fluctuation of the pitch in a sound source signal to be outputted, and to obtain the natural synthesized sound with high quality. |
