Sound signal synthesis method, generative model training method, sound signal synthesis system and program

摘要

The sound signal synthesis method realized by a computer is to combine the pitch data including a plurality of sound name data corresponding to a plurality of sound names into the sound name data specified by the sound name of the sound signal to be synthesized. A hot value is set according to the deviation of the pitch of the sound signal from its name, and the generated model is used to estimate the output data indicating the sound signal according to the set pitch data, and the estimated output. Sound signals are synthesized according to the data.