But it's not important what they (or I) think constitutes accurate, it's important what is accurate.

If a recording of a cello sounds like a real cello then it's accurate.  Even if I perceive a cello to sound different to the way you do, if the sound waves reaching my ears from a system are the same as those from a live cello I will think it sounds like a cello, and so will you.

Now, if we took a recording of a cello and perhaps warmed it up by tweaking up the bass response a little and then played the original and the tweaked version to a group of people, it's quite possible that many would prefer the tweaked version or think it sounded more realistic (or both) but they would be wrong.  Once presented with a real cello playing infront of them they would then realise this.

The trouble is (and it's one of the reasons that subjectivity in audio is such a problem) that human audio memory is extremely poor.  Our ability to remember precisely how something sounds is almost non-existent so we are totally unable to make any accurate judgement about how realistic a system sounds just by listening to it, even if it's playing a recording of a live performance that we may have heard only hours ago.

Michael.