| 摘要 |
PROBLEM TO BE SOLVED: To make a synthesized voice more natural and easy to listen to. SOLUTION: A target text to be read aloud is converted to an intermediate code and inputted to a rule synthesizing process part 5. An intonation pattern is calculated from the intermediate code. Phonemes matching the pattern are selected from a use adaptation type sound source 14 and connected. Here, the sound source 14 is generated from voice data recorded when a sound source generation text in the same use category with the target text. Thus, the sound source matching the use category is used to generate a synthesized voice which is naturally listed to. |
