Disclosed herein are methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a musical work adaptation quality evaluation model. One of the methods includes: obtaining a plurality of adapted musical work samples; separating audio tracks of each of the plurality of adapted musical work samples; determining instrument information identifying instruments played in each of the audio tracks; performing originality analysis on each of the audio tracks to obtain an originality value corresponding to each audio track; for each of the audio tracks, using a combination of the instrument information and the originality value corresponding to the particular audio track of the adapted musical work samples as a sample feature; and training a musical work adaptation quality evaluation model by using sample features of corresponding adapted musical work samples as input, and a predetermined evaluation score of the corresponding adapted musical work sample as output.