| 摘要 |
<P>PROBLEM TO BE SOLVED: To synthesize sentences and singing voices with natural voices close to a human voice. <P>SOLUTION: In the speech synthesis device 200, a singing voice synthesizing part 212 prepares singing voice prosodic data on the basis of a text part analyzed as singing voice data by a tag processing part 212. A language analyzing part 213 applies language processing to a text part other than the singing voice data. As a result of the language processing, as for a part registered in a natural prosodic data dictionary, corresponding natural prosodic data is selected, and a prosodic data adjusting part 222 adjusts a parameter on the basis of phoneme piece data of a phoneme piece storing part 223. Meanwhile, as for a part unregistered in the natural prosodic data dictionary, a voice symbol generating part 214 generates a voice symbol string, and a prosodic data generating part 221 subsequently generates prosodic data. Then, a waveform generating part 224 connects needed phoneme piece data on the basis of the singing voice prosodic data, prosodic data and the natural prosodic data to generate voice waveform data. <P>COPYRIGHT: (C)2003,JPO |
