METHOD AND DEVICE FOR SPEECH SYNTHESIS

摘要

PROBLEM TO BE SOLVED: To provide a method and a device for speech synthesis that provide a natural and satisfactory speech synthesis output by a system having a relatively small DB. SOLUTION: The method includes the stages of: (a) determining and sectioning a speech unit to be connected into a left unit and a right unit; (b) variably determining specified areas to be interpolated as to the left unit and right unit; (c) connecting an extension part to the border part between the left unit and right unit; (d) arraying pitch marks based upon corresponding units at the extension part; and (e) superposing the left unit and right unit one over the other. By this method, speech units are connected by using a DB and acoustic inconsistency at the connected border part is reducible even with the small-capacity DB based upon a corpus, thereby embodying speech synthesis of good quality. COPYRIGHT: (C)2004,JPO&NCIPI