7V, greg and PeteH have it about right.
I do a lot of singing, primarily with chamber choirs, and whenever I get the chance during a rehearsal (when I'm not supposed to be rehearsing) I will sit in as many places as I can in the auditorium, as well my seat in the Tenors, (typically, I'll close my eyes) and listen to the sounds, detail, imaging etc. as we have all learned to do when communing with the spirit of our hifi/HOBBY. And I've sung for recordings and broadcasts as well as the live audience
One of my favourite local auditoriums is the Bridgewater Hall (I perform there about a dozen times a year) and on Sunday morning I was singing live on Radio 4's Sunday Worship from Emannuel Church, Didsbury, listening carefully ... And my son was in the congregation (of 3) before we went off to the Manchester hifi show to listen to lots of hifi

I love seeing what the sound engineers have done with mike placement and listening to the end result.
Anyway, the point(s) of all this. In my opinion/experience...
In a live performance, the brain recognises the distance from the source of sound and compensates its interpretation of the signal so that what it hears sounds "right" for the acoustic it's in. You are always conscious of sound coming from all around you - probably most obvious in a highly reverberent acoustic, eg church with organ/brass. One of the most impressive sounds I ever heard was the 6foot diameter bass drum in the Dies Irae from Verdi's Requiem, in Kings College Chapel, Cambridge - classical rock music; there's no subtlety to what Verdi intended - and with something like a 4 second decay to the sound. Wonderful to hear while you were there, but I don't think there's any way you would want to hear the equivalent recording without the performers being brought up in the mix relative to the reverbaration.
Soloists often disappear beneath the rest of the parts, but part of that is that the accompanists haven't got the balance right (ie. they're too loud) and the other is that that's what the composer intends. On the other hand, why do Wagnerian Sopranos sound like that? because that's the only way they can produce a sound which stands half a chance of getting over the top of the orchestra and filling a large auditorium. Oh, and the vibrato helps you hear them, because it both increases the volume of the sound produced and the ever moving frequency means you don't confuse a single/pure tone with the sound of another instrument.
Probably the only place you'll get a stereo image (breadth, pin-point location) similar to a recording/hifi is if you stand less than 12' behind the conductor (or preferably in place of the conductor). Think about it. In a typical hifi listening postion, the speakers will form an angle of about 30degrees from your listening position. In a concert hall, a full orchestra may only span an angle of 5-10degrees.
So to hifi. Of course the positioning and detail is an important measure of the accuracy of the system. Fundamentally it is the system's job to accurately reproduce the recording, not to compensate for the recording. The ear/brain is capable of determining postion and detail with surprising resolution, and ear/brain will detect the accuracy and phase information provided by a good system. Whether the recorded signal is realistic isn't the system's fault. However, it may also be that "compensating" for the recording will create a more realistic "as live" reproduction
You seldom get as much treble information live as you do on a recording (because you're too far away). Major exception is a symbal. Sound of finger sliding down wire-wound guitar strings? You'll only get that from an acoustic if you stick your head next to the fretboard.
I found it fascinating when listening to the imaging of a friend's Quad II/22 amps how they generated this wonderfully Lilliputian sound stage - a scale model of the band/orchestra in my room. Not realistic in that sense, but still delightful.
Binaural recordings? Some years ago (back in the 70's?) the BBC did some broadcasts with binaural recordings. Fascinating, but for exactly the reasons Uncle A alluded to I suspect they either used binaural for the ambient sounds and close(r) miking for the main sounds, or they were very conscious of just how close the dummy head had to be to the main subject.
One of my biggest disappointments with the (classical) recording process is the lack of emotion, performance and spontaineity on the final recording. You go for the 1st take, give it everything you've got and then get told that there's imprecision on a partivular consonant, or whatever. So by the 5th take and by the time every phrase has been chopped apart you're going for safety and precision and not performance
Moving further off topic, the contrast between what I heard 1st thing on Sunday morning and later that day makes me believe I seldom hear "hifi" that truly sounds like the original instuments. There was a demo of Digital Audio's Hailey speakers which was sadly lacking in music, but was reproducing some surprisingly realistic individual instruments. And the Origin Live system which was described by the rep as "a bit bright" was unlistenable. I've come to believe that the euphamisms "dynamic" "good with rock" mean that the system adds the same distortion as you're used to with stage amplifiers. I've spent years standing next to the brass and percussion sections and wil seldom (have never?) heard in real life the treble distortions that I hear from some of the "best" hifis. The Chord/Spendor system was simply distorted and unrealistic to my ears when I went into their demo room. Most stuff is artifically bright (or maybe... the demon CD :JPS: screws up any chance of realistic music)
Anyway, enough of this rambling. Not as coherent a response as I probably intended - I'm sure there were observations I meant to add that have been omitted and definitely things I mentioned that are of no relevence - but I was up working all last night and brain, vocabulary and fingers are not in perfect synchronisation this evening.
zzzz......