| 摘要 |
<P>PROBLEM TO BE SOLVED: To correctly analyze conversational text. <P>SOLUTION: A voice synthesizer 10 has a correspondence dictionary storage section 13a which correspondingly stores character strings of conversational representations and character strings of literary representations, and a receiving section 15a which receives a first text. The voice synthesizer 10 also has a permutation section 15b which permutes character strings matching a conversational representation in the correspondence dictionary storage section 13a, in character strings of the first text, with a character string of a literary representation corresponding to the conversational representation. The voice synthesizer 10 further includes a morpheme analysis section 15c which performs morpheme analysis on the first text and a second text produced by the permutation of the character string. The voice synthesizer 10 includes a comparison section 15d which compares a likelihood regarding a connection of character strings on the basis of results of morpheme analysis of the first text and the second text. The voice synthesizer 10 further includes a corresponding section 15e in which, when the second text has a higher likelihood in the connection of character strings than the first text, the result of morpheme analysis on the second text is made corresponding to the first text. <P>COPYRIGHT: (C)2012,JPO&INPIT |
