摘要 |
PROBLEM TO BE SOLVED: To make obtainable various extremely natural synthesized voices by using a voice reproducing function and a synthetic voice function at the same time. SOLUTION: A rhythm control part 13 generates rhythm information corresponding to phoneme information based on the phoneme information of non- routine expression contents by a rhythm generation part 14, and adjusts the rhythm information corresponding to the non-routine expression contents by considering consistency with the rhythm information of routine expression contents taken out from a voice data memory 11 with attached information by a rhythm adjustment part 15. Then, by connecting a voice signal corresponding to the routine expression contents with a synthesized voice signal corresponding to the non-routine expression contents by a voice connection part 18, the voice signal of the expression contents containing the non-routine expression contents in a part of the routine expression contents is generated according to input information. |