| 摘要 |
<p>PROBLEM TO BE SOLVED: To decide length of a silent part so as not to become unnatural speech synthesis. SOLUTION: When a value of the length of the silent part L0 between a first phonemic line 'ko' and a second phonemic line 'tode' that are stored in a speech data base 1 is decided, a means to decide the length of the silent part 4 compares a length of a silent part 1r after 'ko' with a length of a silent part L3h before 'tode' that are recorded in the speech database 1. The means to decide the length of the silent part 4 decides L1r that is smaller in value as the best length of silent part L0 and directs it to a waveform data connection part 5.</p> |
