Thanks for all the interesting replies.
It seems the majority here fall in to one of the categories I mentioned in my first post: It's impossible to recreate the real sound (including my particular focus on instrumental timbre), thus I don't expect it of a sound system so I don't care about that goal.
At least that last "therefore I don't expect it and don't seek that" is an inference. But surely, don't many people here care at all that a piano sounds like a piano? A sax like a sax? To *some* degree? And isn't familiarity with the sound of real sounds - e.g. voices - one of the ways we can identify with our ears how a reproduction departs from the real thing? I.e. you could reject a speaker without even measuring it based on how obviously it colored the reproduction of the human voice - an obvious dip in the frequency spectrum, or a resonance giving an artificial "chesty/boxy" coloration. So a residue of my question remains: Given a loudspeaker can't perfectly reproduce the real thing...to the degree it CAN reproduce some aspects or semblance of the real thing, do you not use any comparison to the real thing AT ALL or care?
For me, the single most important characteristics of a "High End" audio listening experience is a timbral presentation that gives me some of what I like about the sound of real life voices and instruments. If the tone/timbre sounds obviously "off" to my ears, I simply have no desire to keep sitting and listening. I don't have these demands for music that I listen to elsewhere - in my car, in my kitchen while cooking, as background music.
I think the idea that one needs high end gear to properly enjoy music is ridiculous. I truly, deeply enjoy even listening to music from my iphone's speakers - select a song, or internet radio station, toss my phone on to my bed while it's playing music as I'm getting dressed or whatever...love it!
So I don't need a high end system in order to enjoy music. But if I want to be compelled by both the music AND the SOUND, which for me is the point of wanting a higher end system, then I prefer a system that can, in some important-to-me-ways, remind me of the real thing.
And for me some systems can do this, others just do not.
For decades I've been habitually closing my eyes when listening to real voices and acoustic instruments to inventory the characteristics that I love so much about it, and which distinguish it from most reproduced sound. Same when I listen to sound systems - why don't I buy this? What characteristics depart from the real thing?
When I listen to acoustic guitar, or play my own, it invokes in my mind certain tonal qualities/colors/characteristics. It was always sort of vexing that acoustic guitar through some systems actually DID seem to recreate the essential tonal colors/characteristics I apprehend in the real thing.
As in "yes, that's right, that's essentially how those sound to me in real life!" While others didn't evoke this sense of "rightness" whatsoever.
I found that "rightness" to be utterly compelling and satisfying.
Of course there's all the problem of bias involved and, like almost all of us, I had no easy way to do some double-blind tests between real instruments and reproduced. But I did still wonder if I was actually carrying around some usefully accurate sense of "the sound of real instruments" that was doing any work when listening to systems. As a very basic touchstone I long ago made recordings of my wife's voice, my kid's voices, instruments I knew well - my acoustic guitar, my son's playing sax and trombone etc. So I could do live-vs-reproduced comparisons
for speakers I had in my home (I used to review for a brief while, but aside from that had many different speakers), and as a touchstone to bring to auditioning speakers. In the close-my-eyes and listen to the live-vs-reproduced sounds, especially when it came to my wife's voice and my acoustic guitar (I'd have a friend play it), there really DID seem to be that difference I thought I was detecting. Some speakers really DID seem to be reproducing the guitar with recognizable timbrel verisimilitude, while others just didn't. For me, discerning a believable timbrel representation between speakers is like viewing different TVs to see if the color seems accurate, especially for well known colors like skin tones. There's always going to be deviations for many reasons in terms of a TV perfectly reproducing skin tones (not least because the sources and photographic techniques/quality/artistic choices will vary), but we can still identify when a TV's color is really "off" when it never renders realistic skin tones.
Frankly I've been amazed at how authentically some speakers have rendered the sound of my acoustic guitar, where it was *almost* effortless to imagine someone was playing my actual guitar right in front of me. Whereas other speakers can produce a very convincing sense that *something* is playing very vividly in front of me...but it's not like the real thing because the tonal "color" is wrong.
And I still use that touchstone: I don't expect any sound system will produce total accuracy, the true range of tonal characteristics of "the real thing." But I want it to generally be able to produce something characteristic of the real thing, something important to the way I hear things.
So acoustic guitar recordings may not be perfect facsimiles of what occurred in front of the microphone, but they *do* generally invoke in my mind what I hear from acoustic guitars, that particular combination of "woody body, sparkly metallic strings" that produce the "tonal colors" I seem to experience from real guitars.
And with this "tone-first" attitude when evaluating sound systems, I've noticed that every speaker system I've encountered seems to homogenize timbres in a specific way. That is, once I hear just a few instances of drum cymbals, or a sax, trumpet, acoustic guitar or whatever on that system, there is no more timbral surprises really. I pretty much know what drum cymbals and those other instruments are going to sound like, in terms of tonal color, from then on. So given every system homogenizes to some degree, I prefer to have a system that at least "homogenizes" to a tonal quality that *generally* or often seems consonant with what I hear from the real thing.
But, again, I remain curious to what degree others here use the sounds of voices or real instruments as any sort of touchstone in choosing speakers, even within the understanding "it will never truly reproduce the real thing."
My sense from reading web sites like this is that most members evaluate mostly on technical parameters off accuracy, and if it measures accurate...well...what else should I bother worrying about?