摘要 |
PROBLEM TO BE SOLVED: To provide a device and a method for efficiently generating a voice synthesis dictionary which matches with the utilization objective of a user and has a suitable size. SOLUTION: A large scale corpus dictionary is constructed by inputting one or more than two keywords and text data and they are made correspond to each other and stored. Then, voice data corresponding to the text data are stored one or more than two second keywords are inputted, text data in which a keyword and the second keyword are matched with each other are extracted from the large scale corpus dictionary based on the second keyword and a new dictionary is generated by extracting the voice corresponding to the text data. |