MP3’s, even 320k CBR true stereo, just … no. They have no place here. The artifacts of lossy compression are obvious even with mediocre gear, quality is bad enough to be disturbing. Anyone advocating hires audio should certainly notice the distinctive “MP3 sound” even without any comparison.
Sometimes Tidal acts up and sends me a “normal” quality stream - I tend to catch it even if music is just playing in the background.
Sample rates get tricky. By and large, CD quality is enough, I can’t spot a difference. 48 kHz may have the slightest edge over 44,1 kHz, beyond that, definitely doesn’t get any better. Sample rate has everything to do with high frequencies; My ears these days probably don’t do much over 18 kHz. Besides, oversampling and such magic is used to counter well-known issues.
Having a choice of bit depths, more seems better, but it’s barely noticeable even when I know which is which (16 vs. 24) - the difference is most likely coming from somewhere else. 16 bits can carry more dynamic range than I can hear.
MQA … I hate everything about MQA (proprietary, licenced, DRM, lossy and totally unnecessary). I’d very much like to say that it sounds worse than pure 44.1/16 even when unfolded at least once, but can’t. Better though? So far, I’m not convinced.
There are also two big problems. First and foremost, all material released with rate/depth above 44.1/16 tends to be of higher overall quality right from the start. We should only ever compare hires audio to that same true hires audio file (not MQA) downsampled to 44.1/16. I’m just as lazy as anyone else and happily keep comparing apples to oranges and speculating.
Second, my environment. I listen mostly via an USB DDC (Topping D10) plugged through an impedance converter into the digital inputs of a pair of Genelec 8330’s and a 7350 sub. I can also use the D10 as a DAC and route the analog output via a balancing line mixer to the analog inputs, but it’s pointless, unless I actually mix several sources. The monitors treat everything as 48/lots-of-bits, be it from AES/EBU or the analog input, which internally first goes into an ADC…
At work I use ok-but-nothing-special headphones (AKG K272), which could render frequencies up to 28 kHz, plugged into an old Lead Audio LA-100 DAC/HPA. Via it’s S/PDIF inputs it can take high sample rates, but it filters out anything over 20 kHz. You’d be surprised how many amps do, for a good reason: it’s the domain of electromagnetic interference and other noise, more than anything else.
So much for those sample rates, then.
In my perfect world, everything would just be standard 48/20 FLAC’s (and everything would support that bit depth too). Not so much more bytes, yet quite comfortably above our human abilities…