Long story short -- 44.1kHz will record up to 22.050kHz just as accurately as 96kHz will.
If you have a source that can reach above that, a mic that can pick it up, processing that can react to it, a monitoring chain that can reproduce it and you're main clients are dogs and bats (that can actually hear it), there you go. Otherwise, if you can't make an award-winning, audiophile celebrated recording at 44.1kHz, the problem isn't your sampling rate. And as a "freak of nature" who at least at one time could hear 22kHz, which made me a sort of "lab rat" for my fascinated audiologist, there is nothing pleasant about that sound anyway. Consider yourself lucky if you cut off at a more "typical" 17-18kHz.
Long story short, 44.1kHz exceeds even exceptional human capabilities. So that moves us on to word length -- I *am* a fan of 24-bit for several reasons up to and including that same reason as before -- It exceeds human capabilities.
That said - Most recordings these days aren't ever going to come close to using that sort of dynamic range -- But again, it exceeds the capabilities of all your gear also. You aren't going to find recording gear that has a >120dB usable dynamic range. Not mics, not preamps, not speakers, amplifiers, compressors, etc.
THAT said, with a "theoretical" 144dB dynamic range (even if most gear won't spec out at more than 120), at least you're capturing
everything that gear is doing, along with a frequency response that exceeds human potential. And self-noise aside, recording in 16 bit can add dither noise -- On one track, not a big deal. On 20 tracks, that can be a big deal (although again, you're talking less noise than a decent tape machine, so it's somewhat insignificant in most cases).
Now on the "But... but..." side -- There are still many who suggest higher sample rates because they think it's picking everything up more accurately. (A) It's not true in the first place. (B) I have had so many recordings in here at high sample rates -- Thank goodness most not over 96k, but some up to 384 (and those are usually the ones that wouldn't matter if they were recorded to a broken cassette deck). A *LOT* of these recordings actually have - what's the Yiddish word - Schmutz? A bunch of garbage above 24kHz or so. Just a lot of noise and crap.
"How did you hear it, John?" Good question -- Play it back at 1/2 speed.
Now, I don't keep track of it, but there were occasions that I asked what sort of converters were being used. Let's just say that no one ever said Lavry, Crane Song, RME, Weiss, Burl or
Prism. But there were a lot of fairly common "budget friendly" names in there. What all that crap is/was, I have no idea. But those same converters seemed fine at normal rates. And again, this wasn't audible schmutz - But if wasn't part of the original signal and it shouldn't have been there. I'd call that a defect that was only present when using 2x rates.
So if you *do* want to record at high rates for whatever reason, I'd be looking for some "nice" converters. That said - You'll probably like those converters so much at normal rates that you won't want to bother.
[EDIT] Don't think I didn't notice that I started this post with "Long story short" and then said it again not a couple paragraphs later...
