I always wonder about industry pros...the ones with the golden ears...making claims about greater clarity and depth when they switch their expensive converter from 48 to 96 to 192.
I mean....sure there's the math, but what are these guys really hearing?
If they use the same converter and just flip their sample rates...what else is different?
And I don't ask that question to initiate a deep debate about sampling rates...but I'm just pointing out why so many people who are not golden ear pros end up considering that notion that more=better.
They are being TOLD it does...by people they respect/admire/believe...etc,.
First of all, just because someone is an "industry pro" does not necessarily mean they have golden ears. I don't mean that as disrespect, it's just a statement of fact.
The number of pros in this racket that truly have golden ears is probably quite small; amd many of them that *did* have golden ears when they started out in analog 40 years ago, probably don't have the same ears now that they used to, just due to a combination of age and extended exposure to sustained sound levels. (Even listening to music at low levels will wear down the hearing over time when done more often than someone who doesn't do it as often.) I'll bet that if everybody told the truth about it, the number of real pros who work with some level of tinnitus in one or more of their ears would be shocking.
And let's not forget that even the pros are human beings, and there is a tendency for audio engineers to be gear sluts. Again, no disrespect, it's a lovable quirk if nothing else. Engineers like their toys. It's like the fisherman who has a zillion rods and reels and lures and flies and so forth, and is always on the lookout for the the latest (new or vintage) k3wl toy. And it's human nature to defend or justify esoterica, even if the justification is not necessarily there.
Add to that the "gear list factor" in recording studios; the documentable fact that half the gear in big studios rarely ever actually gets used, but if they didn't have it in their gear list, they would not attract as many customers. And as John ("Massive") said correctly back in the beginning of this thread, the majority of industry professionals actually do NOT use higher sample rates most of the time.
As far as the question about flipping the switch and hearing a difference, there's a couple of things that often can account for that. The fist is that a given converter may actually sound better at that rate. The second is psychosomatic, which is why truly double-blind testing is the only way to know for sure what one is really hearing. The third is that any given converter design may have a "sweet spot", for lack of a better term. To go back to the car analogy, it's kinda like where an engine's power band may be located, with this engine delivering more power (or maybe being more efficient) at this gear ratio and RPM, but that engine performs better at a different range. As was said near the beginning, some converters may sound better at 44.1 because they just are not designed to work well at 88.2, or it could also be vice versa.
The problem with the "math" of A/D and information theory (which is a whole field that covers a whole lot of stuff that has nothing necessarily to do with music or audio) is that it is very difficult for the non EE/non info theory specialist, and it's not easy even for the specialists. Things like Fourier transforms and the Nyquist theorem and such sound like fairly simple concepts on the surface, and that can be very misleading, because the actual math behind them is way over the heads of most audio engineers, even the pros.
I don't claim to understand it all myself, either. But I understand enough to know that many of the textbook representations and common understandings of how it all actually does work are off the mark. There's a famous saying about quantum theory in science, that says something along the lines of, "there are only five people on this planet that truly understand quantum theory, and four of those are probably faking it." Information theory and the actual science and mechanics of A/D//D/A is kind of like that in that the reality of it is not very intuitive, and that the pictures and concepts we imagine to help us understand are not really right.
The fact remains that the Nyquist theorum states that in order to reproduce a signal at a given frequency within a limited bandwidth signal, you need to have a sample rate of at least twice that frequency. It does NOT state that the higher the frequency, the more accurate the representation at that frequency, it says that at a sample rate of twice the frequency, that frequency is *reproduced*. "Reproduced" here means an exact copy, i.e. a lossless representation (remember this is *information* that Nyquist is really referring to, and any loss of information would not be a reproduction.)
The stair step/connect-the-dots/thinner slice model usually used to teach about sample rate does not really work on a real level because there's no way to connect the dots at only twice the sample rate to make the original waveform with any certainty or accuracy. That model is not really an accurate description of what's actually going in.
Just like with many concepts in high science, such as it is impossible to accelerate a body to a speed faster than the speed of light, unless one can comprehend the math themself - which most folks can't - they're best off just accepting it and moving on to the consequences of it, otherwise they'll just be another one of those crackpots who ignore all the real math and real evidence and write nonsense letters to some scientist claiming that they have a simple alternative to Relativity.
Any difference in sound between sample rates on any given converter is virtually certain to NOT be a function of the actual sample rate itself, because that's what Nyquist shows. Any actual difference heard - real or imagined - will be due to some other factor or factors than the sample rate itself.
G.
