Maybe i can clarify a few points...
1. THE DOG EARS MYTH. While in theory the 192KHz audio spec might seem to suggest that all audio up to 192KHz is recorded, in practice most 192KHz converters employ low pass filters somewhere just over 20KHz to prevent aliasing artifacts (figuring the human ear can't hear up there anyway so nothing is lost), so 192KHz converters generally wouldn't even let your dog hear anything above about 23KHz or so.
So why do 192?
2. THE ALIASING THANG. All converters have to issue a pulse for each clock cycle. That pulse, while very quiet, disturbs the waveform in ways that are not necessarily musical or desirable (noise). 192KHz allows that pulse to be moved way up above the audio spectrum to get *most* of the noise out of the way.
3. THE RESOLUTION THANG. Neil Young, at the height of his anti-digital days, used to rant about square waves. The fact is that, the all-important subjective experience aside for a moment, a waveform at 44.1 *does* look like a staircase compared to the same waveform at 192 (especially at lower signal levels, where resolution deteriorates even further), and 96 is in the middle (Loony Toons?), while DSD (2.8MHz sampling frequency) looks more than twice as smooth as 192. Comparing 44.1 to 192 is like comparing South Park animation to Pixar.... some people might subjectively prefer the choppy animation style of South Park over the silky smooth kinda-realism of Pixar animations, but people who are into "realism" are obviously gonna prefer Pixar. 192 has twice as many "frames" per second as 96, so the digital audio "cartoon" is going to be smoother and more realistic - there's more information there, so it better approximates the original waveform (assuming the waveform hasn't been mangled by processing somewhere along the line).
4. THE TRANSIENT RESPONSE THANG. Back in the real world, one place where 192 (or, better yet, DSD) really makes a difference is with transients (very, very brief sound events). The lower the sample rate, the more distorted the transients, because the *slower* sample rates don't react quickly enough to accurately represent very short-lived events. Since higher frequencies have shorter lives, this transient response is more noticeable at higher frequencies - try listening to a triangle or high-pitched bell recorded clearly at 44.1, 96, 192, and DSD, and see if you can't hear the difference. Are most of you gearslutz recording a lot of triangles these days? I dunno, but personally i like to capture all the shimmer of cymbals and the magical sparkling transients from my Matchless amp, too, through high-end mics and preamps that actually register all those transients, and higher sample rates clearly do a better job of preserving that glorious detail. But if you earn your living recording fart noises through an SM57 (nice work if you can get it), you might or might not notice much of a difference at 192 vs 44.1 or 96, knowamsayn?
5. THE "FEEL" THANG. Some engineers argue compellingly for finding alternatives to the LPF, so as to allow the rest of the spectrum up to 192 to be rendered clearly - not because humans can *hear* above 20something KHz, but because we can *feel* it. Whether one is consciously aware of feeling something different at 192 is an entirely different question, but, hey... waves is waves; there's something physically different going on at 192 - it may be subtle, but it's objectively real - and who's to say someone else doesn't notice or care? It is understandable, then, that both feel-oriented audiophiles and technical purists alike don't wanna mess with even those parts of the spectrum that can't be heard but certainly can be felt (at least by some of us, and i count myself among those).
6. THE... ER... SOUND... THANG. Well, folks, we like to say that, in the end, all that matters is what each of us hears and prefers... and some of us, myself included, do hear a difference at 192, and we prefer it. I also hear the difference between well-recorded DSD and 192, and i prefer DSD - it's indeed the closest digital has come to 2" 16-track. But that's me... YMMV, etc.
Oh, yeah, and before i go, let me address one more myth... someone mentioned the notion that bit depth (e.g. 16 vs. 24) is more important than sampling rate in determining sound quality. Well, this used to be subjectively true, at least to some folks, when all there was in digitaland was PCM audio. The jump from 8 bit to 16 was a quantum leap, and the hop to 24 didn't suck either. But DSD turns all that on its arse... it's *one* bit, yo. One bit moving at just over 2.8 *MegaHertz*. At that speed, that lonely little solo bit has more detail than 24 bits at 192KHz - it's more like good analog tape, because it's sampling the sound nearly 3 billion times a second... a lovely curvaceous waveform for Neil and the gang.... So, in this case, increasing sampling frequency has a much more dramatic benefit than bit depth.
But humans can't hear anything over 20-to-23 KHz, right? Right (usually), but clearly music sounds better at 2.8MHz than it does at 44.1 or even 192KHz (1 bit at 192KHz is roughly equivalent in resolution to 24 bit at 384 KHz), all other things equal. So the superiority of DSD is a prime example of how the whole "we can't hear way up there anyway so what's the point" argument is founded in misunderstanding - presumably originating from non-techies reading techie threads and misunderstanding them and then passing them along to others until the misunderstaning grows to urban legend proportions.
Fellow gearslutz, higher sample rates are not about *hearing higher frequencies*; they're about increased resolution, accuracy, detail, transient response, and subjective and objective realism in the frequency range we *can* hear, and... also stuff that's subtler and harder to express in words (and thus harder to "defend" in jousts of naked, ignorant wit) but nonetheless real and powerful and valuable... ya know... the "energy" of a performance.
Whether a given converter box or signal path actually *delivers* the full picture of purity and energy that the technology theoretically allows for is an engineering issue (and no doubt poor engineering will continue routinely to be paved over by compelling marketing) and, as always, beauty is in the ear of the beholder. But at least let's talk about the real issues with high sample rates as they apply in real world applications (e.g. the unequivocal benefits on shredding triangle solos vs. the negligible improvents for flatulence tracks).
Have any good gear to donate to a great arts-oriented charity for an end-of-year tax deduction? http://www.creationation.org/heart/help.html