摘要 |
PURPOSE:To provide a voice synthesis method in which segmented classification by voice attribute is limitted to only three kinds, i.e., soundless, voiceless and voiced sound and the precision in segmentation is improved. CONSTITUTION:Voices corresponding to the voices to be synthesized are inputted to a voice inputting section 11. The inputted voices are analyzed in a voice analysis.segmentation section 13. The inputted voices are segmented in accordance with their voice attributes and rhythm characteristics of the segmented voices are extracted in a rhythm characteristic extraction section 14. Based on the extracted rhythm characteristics, the time duration of synthesized voices, that are set by a rule, power and other rhythm characteristics are changed and the voices which have desired rhythms are synthesized. |