摘要 |
<P>PROBLEM TO BE SOLVED: To provide a method, a device, and a program for speech synthesis that can use a cost function taking into consideration properties of words by search words of a database for speech synthesis. <P>SOLUTION: The speech synthesizing device 1 which synthesizes a speech of text data is equipped with a word division/phoneme and rhythm information addition part 5 which divides the text data into words and adding phoneme and rhythm information as information including phonemes, rhythms and basic frequencies to the words, a data search part 7 for speech synthesis which searching the database 13 for speech synthesis for word division candidates having minimum phoneme and rhythm costs and connection costs according to matching of words in the database 13 for speech synthesis against phoneme representation by a dynamic programming method, a phoneme and rhythm correction part 9 which makes corrections at connection parts by word division candidates of a searched speech data sequence and at parts different from predicted rhythm, and an output part 11 which outputs correction results as speech synthesis data. <P>COPYRIGHT: (C)2004,JPO |