This is another really great video that I humbly think EVERYONE who works with digital audio should check out: There are no "stair steps" in digital audio ! What The Matrix can teach us about “resolution” - Production Advice
I work with 88,200 at all stages until the master when I bring it down to 44,100. When doing this process however, never forget to add dither, otherwise you will hear added artefacts that you do not want!
Working here with cubase 10.5 and processing vocals a lot and in detail using variaudio I go for 192kHz or 176.4kHz or 96kHz for recording and editing vocal projetcs. If working in 48 or 44.1kHz I hear much more artefacts and can't stretch audio as much. Meanwhile for regular instrumental projects that don't need intense audio manipulations I often work in 48kHz. Regarding bit depth I recently switched from 32bit float to 64bit float mainly because in my case it's a bit lighter for my CPU.
Higher precision = less mistakes = more congruence [less dissipation (blurring)] = higher resolution.
And? Where fps here? I don't know about video stuff, dependency of video fps on audio sample rates etc. Don't know. But why can't you work with 48 or 96? Range is another, but speed not. Everything can be SRCed without loss in quality.
In theory, 44.1k samples per sec are perfectly sufficient to completely digitize an analog waveform. However, it should be borne in mind that the digital signal is then manipulated during the mixing process. To prevent aliasing artifacts, for example, the material must either be sampled up or oversampled within a plugin. Information is therefore calculated which is not available, then processed and then degraded again (partly dithered). This up-, down-, up-, down-sample does much more harm than good things. Recording at 88.2k, mixing at 88.2k, mastering at 88.2k, degrading to Redbook 44.1k. Done.
Which are completely reduced to white noise by using dithering. Then the "big difference" is the amount of signal to noise you end up with.
If you work with distortion, saturators, exciters, equing highs - yes, you may use extra oversampling, but not too much. If gentle saturation - no oversampling or x2 oversampling for safety. If insane, x4 or x8 usually 99% enough. Don't overdo. Less is more. Also, as example, Using x4 oversampling at 96k is better than using x8 oversampling at 48k/44.1k. It means, oversampling at higher sample rates leads to less phase distortions or ringing artefacts, they will be far from perceived area, in ultrasound area. It is better to distort 35000-48000 Hz at 96kHz sample rate, than 18000-24000 at 48kHz.
You are a real gold mine ! Thank you Mild Pump Milk for your help and to share your knowledge with others
If you record in 24 bit the big difference is there (compared to recording the same thing in 16-bit) to stay - that is what I was talking about. You talk irrelevantly about what happens when one is reducing the bit depth. But even that is not what happens at all. "Completely reduced to white noise" is a big overstatement. That would mean that re-encoded 16-bit and the original 24-bit file remain (exactly) the same /with some - almost impossible to hear - additional noise/, which is impossible. It just reduces the difference between the recordings to some degree (you can still hear the difference in quality quite distinctly).
Hmmm. That's actually something I hadn't considered. That does make a certain sense... Has anyone done any real comparisons of audio strteching at different sample rates I could check out?
What is "extra oversampling"? What is "not too much"? What does it mean x4 or x8 usually "99% enough"? All this seems very vague, and does not seem to help at all. And the examples given are related to quite a different problem, how oversampling works while processing different sampling rates. so they are irrelevant for your prior thesis.