AUDIO SIGNAL GENERATION METHOD, ESTIMATION MODEL TRAINING METHOD, AUDIO SIGNAL GENERATION SYSTEM, AND PROGRAM

摘要

An audio signal generation system according to the present invention generates an audio signal according to musical score data that expresses each continuous length of a plurality of musical notes and expresses a shortening instruction for shortening a continuous length of a specific musical note from among the plurality of musical notes. Specifically, the audio signal generation system inputs, into a first estimation model, condition data expressing a sound-generation condition specified by the musical score data in relation to a specific musical note and thereby generates a shortening rate expressing the extent to which the continuous length of the specific musical note is shortened, generates, from the musical score data, control data which expresses a sound-generating condition and in which is reflected the shortening of the continuous length of the specific musical note according to the shortening rate, and generates an audio signal according to the control data.