摘要 |
PROBLEM TO BE SOLVED: To provide a voice synthesis method capable of remarkably reducing data capacity, processing amount and exactly reading text information according to an optimal accent. SOLUTION: The voice synthesis method is constituted by providing a character information input part to input the text information, a character information analyzing part 1 to output representation information and accent information by analyzing the inputted text, a voice database 3 to preliminarily store a speech element with plural pieces of interval information, a voice synthesis part to generate synthetic voice by taking out the speech element stored in the voice database 3 and connecting the speech element with the voice synthesis part. The speech element to be taken out from the voice database 3 is defined as the one to be coincident with or similar to the representation information and the accent information as output of the character information analyzing part 1. |