I'm going to go a bit more philosophical here if I may. Apologies.
I think that this question of realism is the traditional view on hi-fi. However, I think that it only made sense when the primary music being recorded was instrumental. However, the advent of electronic- and computer-generated audio, nevermind whether you like it aesthetically, reveals the whole question of "realism" to be a bit of a sham. In this case, the artist manipulates parameters to produce a sound of their liking, regardless of any resemblance to a sound produced by voice or instrument. So, for a song that is 100% computer-generated (again, set aside your feelings about such music, as they are irrelevant here), how does one assess the "realism" of the reproduction when you play the song on your system? And what if one of the sounds in the song attempts to synthesize a traditional instrument sound (say, synthesized piano). When you judge the "realism" of the playback, are you judging how that piano sounds relative to a real piano or are you assessing the perceived realism relative to the information encoded in the recording? For the former, your system has nothing to do with it because it's all dependent on pre-recording decisions. For the latter, well...I have no idea what it even means in the case of synthesized sounds.