摘要 |
PROBLEM TO BE SOLVED: To provide a voice synthesis system capable of reducing the processing load and reducing a compression distortion. SOLUTION: In synthesizing a voice, representative waveforms, waveforms for correcting an error and adaptive time ranges corresponding to a voice unit are read out from a storage device for every voice unit of a given voice unit series. In the adaptive time range corresponding to the first waveform for correcting an error, a waveform string synchronized with pitch cycles in pertinent adaptive time ranges in voice units of the given voice unit series is generated by synthesizing waveforms in which the waveform for correcting an error is multiplied by coefficients in accordance with time from the representative waveforms to the waveform. |