| 摘要 |
PROBLEM TO BE SOLVED: To provide a voice distribution system etc., for speedily distributing information that differs by a plurality of receiving terminals in voice. SOLUTION: When message data, an area ID, etc., are supplied from a message distributing device T etc., a language processing part 1 decides whether the message that the message data represent is an object of speech synthesis and when the message is the object of speech synthesis, a speech segment editing part 4 searches a speech segment database 6 for speech segment data of a speech segment, whose reading matches that of a speech segment in the message and performs conversion of a speaking voice, etc. The speech segment editing part 4 also perform prosodic prediction of the message and selects speech segment data which best match respective speech segments in the message from among searched speech segment data, one by one, according to a prosodic prediction result. As for speech segments which cannot be selected, a sound processing part 31 is made to supply waveform data representing waveforms by unit speeches. Then the selected speech segment data and waveform data that the sound processing part 31 is made to supply are combined to reproduce a synthesized speech. COPYRIGHT: (C)2007,JPO&INPIT
|
