Methods and devices to use and produce a duplicate sound for multilingual text-to-speech synthesis

摘要

(57) by performing processing for sound selectively assembled dual sound for [Abstract] text-to-speech conversion, was selected (diphone), and stores continuously duplicate the sound of each duplicate sound is generated. Dual tone Finally, the treated respectively are stored address to allow individually. In particular, in the inside of said set of which is pronounced by a single person, the dual tones are assigned to different languages ​​or one based on the homophone. By sharing the dual sound selected language among more than one, it is possible to reduce the required storage capacity. Memory may be accompanied by processing modifier language specific.