摘要 |
<p><P>PROBLEM TO BE SOLVED: To obtain a speech synthesis system and a method for creating a synthesis unit dictionary, which are capable of generating a synthetic speech of high quality. <P>SOLUTION: The speech synthesis system is provided with synthesis unit dictionary creation means including: a speech database where synthesis units are stored; first synthesis unit selection means which selects synthesis units from the speech database on the basis of prosodic information for learning; first selection frequency measuring means which measures selection frequencies of synthesis units selected by the first synthesis unit selection means; category division means which divides parameter distributions of synthesis units in the speech database into prescribed categories; synthesis unit extraction means which extracts, per category, a prescribed number of synthesis units on the basis of the selection frequencies; and representative synthesis unit generation means which generates, per category, representative synthesis units from the extracted synthesis units. <P>COPYRIGHT: (C)2012,JPO&INPIT</p> |