| 摘要 |
PROBLEM TO BE SOLVED: To synthesize a voice capable of reducing unfamiliar auditory sense of a phoneme and capable of bringing a natural phoneme on auditory sense, in a connection part, when connecting an elementary speech unit data. SOLUTION: In this elementary speech unit connector 1, a frequency characteristic calculating part 17 calculates a frequency characteristic of a connection portion between connecting elementary speech units, when connecting the fellow elementary speech units, a spectral envelope calculating part 18 calculates a spectral envelope thereof, and a spectral envelope calculating part 19 calculates a spectral envelope of a phoneme environment fit to the connecting elementary speech units. A spectral envelope correcting part 20 corrects the frequency characteristic of a connection portion by the spectral envelopes. A voice waveform duplicate addition part 22 duplicate-adds the frequency characteristic corrected by the spectral envelope correcting part 20, to the connecting elementary speech unit, and an elementary speech unit connecting part 23 connects the fellow elementary speech units obtained by the voice waveform duplicate addition part 22. COPYRIGHT: (C)2010,JPO&INPIT |
