摘要 |
PROBLEM TO BE SOLVED: To provide a speech synthesizer capable of preventing deterioration in naturalness such that a synthesized speech is unnaturally heard. SOLUTION: The speech synthesizer is equipped with a text data analyzing means 41 of analyzing a morpheme of input text data and converting the data into a phoneme with an accent, a phoneme clustering means 42 of clustering phonemes based upon precedent and following phoneme environments, a phoneme accent clustering means 43 of clustering the output from the phoneme clustering means 42 based upon precedent and following accent environments, a speech data search means 44 of calculating connection scores of speech data strings obtained by combining speech data corresponding to the output of the phoneme accent clustering means 43 through a Viterbi search and searching a speech database in which speech data corresponding to phenemes with accents are preclassified and registered for a speech data string having the maximum connection score, and a speech data connecting means 46 of connecting speech data strings. COPYRIGHT: (C)2007,JPO&INPIT
|