It's a complicated question as folks here clearly already know.
Depends on what you are trying to reproduce as well - voices? A big band? A whole symphony? A rock band?
I've always been fascinated by the whole real vs reproduced question so
for many years I even went to the lengths of recording sounds I'm familiar with (I work in film sound) which I would play back through systems. For instance I recorded my acoustic guitar, my son playing sax, my family's voices, other acoustic sounds, etc. Then when I had a speaker in my house I could directly compare the playback of the sound to the real thing. It was always illuminating. Usually those speakers whose "voice" sounded "right" to me in the store or wherever I first heard them, were the ones that passed this test best.
Of all the speakers I tested I'd say the most astonishingly life like (within their frequency range) has been the MBL speakers. I'd had some startling experiences with the 101Ds and Es before - when set up well not only did they manage to produce the most individualized and authentic range of instrumental timbre I've heard, but also the most realistic presence (in terms of bringing objects to life dimensionally and dynamically).
These impressions continued when I managed to buy a pair of the smaller MBL 121 omni-directional monitors. Playing my recordings through those things can be disarmingly realistic. And when I do the "from the other room" test, for instance play the recording of me practicing my guitar, or my son practicing sax, it simply sounds like someone in there playing an instrument. I would not know it wasn't my son in there playing saxophone if you didn't tell me, and in fact I've fooled a few people doing this.
Just a couple weeks ago I played some cuts from Requium For A Pink Moon - Nick Drake music done in an Elizabethan style. This has some of the most natural voice recording you'll find:
http://www.allmusic.com/album/requiem-for-a-pink-moon-an-elizabethan-tribute-to-nick-drake-mw0002357...Played through the MBLs, they recreate the sense of 3 dimensional space, a 3 dimensional performer, with just the right richness, timbre and organic quality, that, when I close my eyes, makes it almost effortless to think I'm at the live performance.
My Thiel 3.7s do spectacularly well with this recording as well, though edged out by the MBLs.
But of course when we start talking of larger demands, we need much more fire-power. I'd think a proper horn set up (lots of them and big) could come closest to reproducing a full orchestra or big band blasting away.
Though, I highly doubt that same system would reproduce voices. They may produce vocals with presence and clarity, but that's not the same as "how voices sound in real life, when someone real is in front of you." And thus far the omnis and certain cone speakers do it best I've heard.
(Electrostatics, like the quad ESL 63s I owned, and the 57s which I love even more, do the startling clarity thing, and get close, but lack that last bit of roundness, thereness and body to vocals to cross that barrier to 'real.').
That's my take, anyway.
And of course there is the age old audiophile question of whether we want realism in terms of "they are here" or "I am there." That is, either the sense of musicians having been transported into our room, or our having been transported to the acoustic event, even artificial ones.
I've had speakers that lean either way, and in fact I'm comparing two speakers right now where one brings musicians into the room and in that sense sounds "more real," the other turns my room into whatever sonic event is depicted.
