Interesting replies. My audio engineering life has been with making amplification as transparent as possible. The more transparent the chain the greater the digital differences that can be heard. So I test out my work with music I have on a few discs that I know very well. Some redbook and some SACD.
For me, percussion and cymbals are the most telling of detail resolution. Shakers, gourds, cymbals have very complex non repetitive overtones that identify them as the instruments they are. Cymbals have sub-tones in the lower midrange that get lost in mixes, intentionally or otherwise. Since I’m also a professional musician (bass) I know their sounds from performing with them.
Orchestral violin blends are also difficult resolution tests. There are harmonics and sub-tones generated in performance that don’t do well in the digital domain at 44/16. But a jump to 24 bits helps and go to 88 or 96Khz and there is great relief from anticipating a colossal fight in the upper 2 octaves in complex passages.
I confess my CD collection was 99% issued before the turn of the millennium. Many are unlistenable on good gear other than for the music content. However some SACDs I have are truly enjoyable. RCA (Sony) reissue of Van Cliburn performances from 1958 and 1960 mastered from original tapes with no processing are a go to for me. First off it’s Van Cliburn. Second, it’s tape direct to disc 3 channel. The inner detail of orchestral passages is always surprising and a great relief to hear. Strings don’t fight with each other. I can hear tape artifacts, while not musical, are less unnatural and on the whole more enjoyable. It’s amazing what was done on an Ampex 3 track in 1958 with Neumann mics.
I also make recordings on a machine that records in 44.1/24 or 16. I hear a difference and 24 is better, no question. Much easier to mix, too. When I had a studio in NYC and recorded a barbershop quartet in 44.1/16 there were some chords that would produce some horrible digital artifacts, so much so that I actually had to mix them in the analogue domain.
Every opinion voiced here is shaped by the equipment the writers use. Some is more transparent than others. Some more forgiving than others. And some ears, better than others.
Which brings me to: EVERYTHING IS A FILTER. Cables, power source, amps, preamps, signal sources, cds, ADCs, DACs, records, streaming, speakers (huge filters), switches, tubes, transistors, op amps, tape machines, mixing boards, virtual mixing boards, rooms, the air in your room, florescent lights, led lighting, ear wax, aging, fatigue . . . Everything affects or alters perception.
Here’s another thing to consider. Upsampling is guesswork. Upsampling generates data that can only be a guess as to what happened between samples. Filters are, too. You pick the nicest sounding guesswork if granted that choice.
For easy math consider this: 44.1khz gives 2 samples of a 22khz event, 4 samples of a 11khz event, 8 samples of a 5.5khz event. There’s a lot of guess work needed to approximate the truth. That said, the brain does its best to fill in the missing pieces. That topic requires a treatise to explore.
My theory of listening and driving fatigue is that the mind is hungry for information. Like driving at night in fog, it's stressful because knowing where the road is is critical to survival. When listening, if you are used to hearing live un-amplified sound or great sound (hi fi) rich with detail and placement cues, the mind notices and strains to hear what is missing in lesser quality recordings or strains to filter out that which shouldn't be there, like harsh sibilants.
So a true 24 bit/ 44.1K recording, not from previously upsampled 16bit, reduces guess work and contains more truth. Double the sample rate and I would concur it becomes harder to discern differences with recordings with sample rates & bit depth above that. Harder but possible.
That said, I can hear file compression artifacts and I don’t use mp-anything for critical listening, AAC is less annoying but noticeable. Youtube is a crapshoot.
So answering the question that this thread is about, I’ll take 96/24 as a minimum for low anxiety critical listening, all remaining things being maximized to get out of the way.
That’s my $.02. Whatever happened to the cent sign that was on my Smith-Corona?
