quote: | Originally posted by Woony
Dude, 96k is nonsense for listening purposes. Anything above 44.1/48k and 16 bit lossless doesn't actually change fidelity, just the noisefloor, which is completely imperceptible with digital anyways.
I started buying vinyl around 2011. I'm at a bit over a thousand now, I think. I haven't been buying that much in the past year for various reasons but i'll get back into it eventually. |
Actually this is not entirely true.
Let me preface by stating I'm a pro mix engineer for the score industry and regularly work in 24b / 96k in true surround 5 or 7.1 and have even worked in 12.1
For most home/consumer listening purposes 16/44.1 is fine. To take advantage of higher sample rates (more on bit depth later) you A) need to have a trained ear B) very serious sound reproduction equipment that doesn;t have a weak link in the entire signal path and C) a good listening environment (i.e. treated room).
All these things being true and equal, you can hear a difference. If one of these are lacking, the extra disk space that 96k takes up isn't worth it
However, there's strong evidence to show that even through the nyquist theorem is correct up to a point, there's harmonics (of high strings, woods etc) that don't get captured in 44.1k becuase they're above the 20k threshold.
With these sounds, you don't directly "hear" them but they add something in terms of being complete with their lower harmonic counterparts that your brain can detect (and inversely does detect if they're not there).
For a real world instance. I worked on a major superhero blockbuster and one of the cues had in excess of 580 individual tracks. Even on two Euphonix 246 channel desks (in different rooms but synced) it took over 30 passes to mix down the cue from nearly 80 stems. In the end there were whole stems of orchestra sections (i.e. complete orchestra sections of say violins and violas) that were little more than "coloration" as you couldn't hear them distinctly but their presence missing did take something away.
The entire score was recorded at least in 96k 32 bit floating point, and in certain instances 192k.
When playing the source vs 44.1k bounce (and bear in mind the bounces were actually passes done on the most expensive equipment and signal chain on this planet so no introduced loss) you could hear the difference like night and day.
Now, if we're talking dance music, made in someone's bedroom, produced from crappy samples and plugins, then mastered with Ozone or even send to a semi pro mastering facility, then pressed to vinyl in some random press, then you're ripping that, I very must doubt there's any real work benefit of 96k vs 44.1k
But as you go up the quality chain, and especially with some older records that were actually made with decent outboard kit, possibly recorded with traditional techniques using mics on synths, or well recorded drums and strings as samples done in pro studios, and pressed at plants that knew what they were doing and had good oldschool engineers, then the argument for higher sample rate (obviously all others being equal as above) becomes more relevant.
Finally, I'm not sure what you mean about it just increasing noise floor. Sample rate nor bit depth increases noise floor by itself. Poor gain staging with higher bit depth yes, but that's a completely different cause and argument.
To jacks point, in most shitty club settings 320 vs .wav won't make much of a difference at all, but the bigger the system, and anything of decent quality (Funktion 1 or martin for instance) .wav will make a massive difference. The problem with mp3 is how the common compression codecs work and the issue for me is that the extremes of LF and HF are always compromised first. I can actually hear the difference reliably between 320 and wav on a system such as my car, and while I don't go to clubs chin scratching, trying to catch the 320 out, you can hear a clarity, and depth of bass with a PCM or other lossless format.
|