Only a fool would by 24/192 "hi-res" files. It's placebo.
Yes and no. While I'm averse to audiophile buffoonery, and there's plenty of it, there are discernible advantages to this format.
The tl;dr is that higher bitrates contribute to exponentially wider amplitude range and the higher frequency sampling mitigates frequency response effects at the highest end of the range of human hearing. The former probably provides, to the average ear, the most readily (glaringly not wishy-washy) discernible difference... but they go a bit hand in hand.
The longer explanation:
First let's get one thing out of the way. I absolutely and completely stand by most scientific findings (Audio Engineering Society etc.) that show that 256Kbps AAC is indiscernible from 16-bit Linear Stereo PCM (your CD audio format, aka CD-DA, CD Digital Audio, "Red Book"). This is simply not open to debate unless controlled, double blind ABX testing (and not conducted over the internet message forums) show overwhelmingly otherwise... and they don't.
On Amplitude Dynamics:
Now, 16-bit Stereo LPCM has 2^16 or 65,536 possible amplitude values per quantization interval/sample. What this translates to is a dynamic range of ~96.7dB. dB is a logarithmic scale in which every 3dB represents a doubling of wave power. What dynamic range represents is range of amplitude levels a system can reproduce from softest to loudest. The absolute values, the floor and ceiling, could vary depending on the mix but in principle this represents the distance from softest to loudest that CD audio, and likewise AAC perceptual coding, can reproduce.
The drawback is that in the last 30 years, the A-weighted average loudness, Leq(A), of popular sound recordings (anything released in any genre to wide distribution, not limited to "pop music"
per se), has increased substantially.
What this means for the listener is that the dynamic range of CD audio is mostly wasted. It also explains some of the obsession with vinyl...
It's not that vinyl is a superior medium or that analogue recordings are better. They aren't. at 80dB dynamic range, vinyl sucks. With its high noise floor and other artifaction, analogue sucks. However, the art of engineering and mastering vinyl properly led to some innovative techniques for "sweetening" a very dynamic mix but keeping the average loudness well within the limits of what vinyl could handle. If the master recordings were directly transferred to digital instead of remastered to obnoxiously higher levels, the digital reproduction would sound flawless while the vinyl pressing would degrade.
An improperly mastered sound recording that peaks above -0dBFS (zero decibels full scale; or the maximum threshold of a given sound reproduction medium before amplitude clipping occurs, which will create distortion at any volume level but becomes profoundly worse to your ears as you increase your system's output volume... A properly mastered sound recording can sound fantastic on just about any stereo system, even the crappy one in my Honda.
But back to the 24-bit argument... 2^24 = 16.78 million possible amplitude values per quantization interval. This translates to a dynamic range of ~140dB. That's beyond the threshold of human hearing, and so substantially greater than 96.7dB.... What it sounds like to your ears is a clearer ability to distinguish even quieter sounds amongst even louder sounds, relative to a CD or vinyl pressing. Imagine true HDR photography for your ears... when you hear the full gamut for the first time, it's rather startling. But also certain sounds like cymbals, with their erratic/spastic amplitude changes, have much cleaner definition to them.
So here we're not talking about a case of two audiophiles fighting it out over some perceived difference that almost nobody can hear and is highly suspect of placebo effect.... If used to its full potential, rather than throwing on Metallica records which have the worst mastering known in human history (I used "Death Magnetic" as an example of totally flat, totally distorted, absolutely terrible mastering in a video I did on the Loudness Wars), it produces a substantially different result from 16-bit.
Now, about frequency...
Sampling frequency, as you know, has to do with being able to produce sounds in the A-weighted range (the range of human hearing, which peaks roughly around 22kHz, though most people have a steep falloff of hearing perception around 17-18kHz). The Nyquist Theorem, developed by Bell Labs in the 1920s, served as the root for determining what the minimum sampling limit would be for reproducing every frequency within the range of human hearing. That is, if the desired range is up to 22.05kHz, then the Nyquist limit or minimum sampling frequency has to be 44.1kHz, enough to minimally represent the peak and the trough of one cycle at that frequency.
When the Nyquist limit leaves little headroom, frequency response roll off and frequency aliasing can occur. But a 20kHz lowpass filter can act as an anti-alias filter. This is, however, not as optimal as simply raising the nyquist limit so that all perceived frequencies are WELL within the system's ability to reproduce with substantial definition and therefore eliminates the need to use lowpass and dithering (1kHz noise) as a substitute.
Again, the rule is still garbage in, garbage out.... So if you start with a master recording that has an average loudness of like... -9dBFS and peaks that peg 0dBFS, versus say an Ahmad Jamal recording from the 1970's like "Awakening" at -22dBFS and peaks below -9dBFS, then the added headroom and definition are totally wasted.
But given some of the standards Apple put in place for the "Mastered for iTunes" label, including recommending peaks no higher than -1dBFS to ensure that the loudest sounds do not clip/distort, there certainly could be a market for some very well-mastered sound recordings and a niche of fans who want to hear them.