Text set creation manner for utterance, text set creation device for utterance and text set creation program for utterance

摘要

P To provide a voice text set creating technique capable of generating a voice text set for optimizing the coverage of phoneme string and prosodic features even when pronounced with a tone other than reading tone. PSOLUTION: A parameter distribution conversion function which converts a parameter distribution obtained from natural voice data of reading tone into a parameter distribution obtained from natural voice data of a target X-tone is previously stored; voice synthesis processing using voice text set candidates is executed by a voice synthesis PROGRAM to obtain a predetermined parameter from voice synthesis data; the obtained parameter distribution is converted by using the parameter distribution conversion function; and the voice text set candidates are evaluated by using the converted parameter distribution. PCOPYRIGHT: (C)2012 and JPO& INPIT