I simply meant - as you mentioned in the rest of your post - that there is no method that can decide for you what is more "realistic" overall. Although perhaps with AI?
I was thinking about all this because I actually went to audition some speakers this morning... It was fun.
While it is impossible to summarize the sound that is heard through a simple set of criteria, I find it equally difficult to listen to recordings and ask myself whether they sound like live music or not, because they simply don't. The dealer played a Johnny Cash track and I closed my eyes and simply enjoyed listening, but I don't know what Johnny Cash really sounded like. I played that David Watkin Bach solo recording which was mentioned in a thread here, and it sounded pretty good. It did raise some questions, but the acoustics were not ideal. I am not that familiar with the track, it is played on an old instrument - it is not so easy to compare it to a reference.
During that audition, I did try to evaluate some aspect of the sound, in comparison to what I am used to listening at home: resolution, neutrality, dynamics/speed., etc.. What did I conclude from it? That I will be returning next week for another listening session!