SINGING VOICE SYNTHESIS SYSTEM ACCOUNTING FOR TONE ALTERATION AND SINGING VOICE SYNTHESIS METHOD ACCOUNTING FOR TONE ALTERATION

摘要

Provided is a singing voice synthesis system that accounts for tone alteration, which is capable of accounting even for tone alterations in singer voice-synthesized singing, in addition to changes in pitch and volume in a user singing voice. A spectral transform curved surface generation unit (119) generates a spectral transform curved surface in accordance with a spectral transform curve that is estimated by a second spectral transform curve estimation unit (117). A synthetic sound signal generation unit (121) transforms a reference spectral envelope on the basis of the spectral transform curve surface in each time point, generating a transformed spectral envelope. A synthesized singing voice sound signal is then generated that accounts for the changes in the tone of an inputted singing voice on the basis of the transformed spectral envelope and a base frequency that is included in reference singing voice sound source data.