The accuracy of the conclusion depends on how the computers played the sonatas. I would hope they were programmed to play with variance and imperfections, otherwise it’s more like comparing beginning musicians’ rote performances to professionals’ emotive ones.