Stop oversampling everything! (plus, higher sample rates have a reason to be)

[pixel85]

I wouldn't put artefacts like background noise on the same shelf as aliasing. Former is usually natural to the ear and often can be pleasant and beneficial (especially white noise).

Most aliasing will probably happen with a lot of sound "content", not just a synth playing sine waves alone (for example), so, it will rather be perceived as additional "smudge", artefact-y whatever. You could say noise.

Are you saying that aliasing is the same as background (white) noise from tape or a room? For me, aliasing (smudge) is closer to distortion than background noise.

btw
While most of the time I perceive aliasing sound as 'lofi' and I don't like it that much, I like it on old Nord Leads, JP8000 or JV1080. Some devices are meant to alias

[ampetrosillo]

If aliasing is not that significant a factor (and I tend to agree), then why bother at all? My point is, if you're bothered about aliasing, then you should embrace higher sample rates, otherwise oversampling is just a partial fix.

[onerob]

There's the Dan Worrall argument about high sample rates and 'intermodulation distortion', which is different to aliasing. He suggests working at 48kHz and using oversampling in plugins, where possible.

[chk071]

He suggests working at 48kHz and using oversampling in plugins, where possible.

Cool. I'm already doing that.

[Ploki]

If aliasing is not that significant a factor (and I tend to agree), then why bother at all? My point is, if you're bothered about aliasing, then you should embrace higher sample rates, otherwise oversampling is just a partial fix.

Oversampling is not a partial fix, and you can’t really get away without it at higher sampling rates either if you push it hard.
A perfect square has enough harmonics that would mirror even at 192k, it needs to be antialiased regardless of sampling rate used

[Everglide]

can reduced phase distortion: the brickwall filter close to Nyquist will either be smoother (so easier to design with less noise and distortion, if converters run natively at the target sample rate, which is rarely the case today, since it's mostly ultra-high speed converters that are essentially sample binning to achieve the final codification) or they will be moved farther away from the audible region.

Can we elaborate on “ultra-high speed converters that are essentially sample binning to achieve the final codification”

Genuinely interested in unpacking this comment further.

[Ploki]

can reduced phase distortion: the brickwall filter close to Nyquist will either be smoother (so easier to design with less noise and distortion, if converters run natively at the target sample rate, which is rarely the case today, since it's mostly ultra-high speed converters that are essentially sample binning to achieve the final codification) or they will be moved farther away from the audible region.

Can we elaborate on “ultra-high speed converters that are essentially sample binning to achieve the final codification”

Genuinely interested in unpacking this comment further.

He’s talking about converter design.
https://en.m.wikipedia.org/wiki/Delta-sigma_modulation
The other approach is “ladder”, mostly reserved for highend boutique converters. (Iirc Lavry and similar)

[fese]

There is a nice quote from Mitchell Froom, which, while not about aliasing and sample rates per se, applies to this discussion as well:
“The difference in emotional effect between a track recorded digitally or analog, on a Neve or SSL, or with this mic or that mic is nil”.

Still, it’s of course fun discussing this stuff, so my opinion is: it very much depends on which plugins you use. Some don’t upsample, some EQs cramp at lower SR (technically they still do at 96Khz), some generally run at 4x upsampling, so running at 96Khz makes no difference. Also, how you use them. There might be a plugin that aliases like hell in plugin doctor, but if you use it on a signal without much hi frequency content and don’t drive it too hard, it might just not matter.
As always in life, it’s not just black and white.

[astralprojection]

Question for you all,
If the label I work for asks for the final product to be in 44.1khz, what would you mix/master at? Curious to hear what you all would do.

i first upsample the wav from whatever samplerate it is in, to 192k. I use rx8.
Then i master it in 192k in cubase using mostly fabfilter stuff. Then at the end, i resample it back to 44k also using rx8. you can ofc skip using rx8 and use any old resampler, but i find rx8 to have really good algorithms and dithering options. the resampler in cubase isnt as transparent, although good enough for sure..

that being said youll be fine using your daws native resampler and just resample to 96k or even 88k. 192k and rx8 is a bit overkill. personally why i use 192k is i think my DAC sounds the best at that resolution thus allowing me to make better decisions while also having plugins running at a very high sample rate so i dont care about oversampling. but i still use oversampling for the compressor and the limiter, even tho im running at 192k native. but then again im a bit of a ocd freak

[martinjuenke]

I noticed that sometimes synths sound brighter and clearer at 96K. I did a 44k/96k mix down recently on a project, all compressors, VST instruments, EQ's everything, and definitely the 96k project was more relaxed, clearer and cleaner sounding. It's not just aliasing, I noticed some plugins just have a different sound and behavior at 96K, like some distortion plugins for example.

Exactly my experience as well.

[martinjuenke]

And yes, the difference in clarity and high range is audible.

Yes, indeed.

[mystran]

can reduced phase distortion: the brickwall filter close to Nyquist will either be smoother (so easier to design with less noise and distortion, if converters run natively at the target sample rate, which is rarely the case today, since it's mostly ultra-high speed converters that are essentially sample binning to achieve the final codification) or they will be moved farther away from the audible region.

Can we elaborate on “ultra-high speed converters that are essentially sample binning to achieve the final codification”

It get really really technical really really fast, but basically the idea is that when we resample the total audio to a much higher rate, we can then quantize it into a few bits with aggressive noise shaping that puts all the quantization distortion into the high frequencies above the audible range.

It's basically the exact same idea as regular dither with noise shaping, but because the high oversampling rate gives us plenty of inaudible bandwidth to dump all the loud noise, we can increase the dynamic range in the actual audible band to the point where only a few bits are necessary and the noise is then removed after conversion by the analog anti-imaging filter that we kinda need either way.

ps. oops, I forgot to check the date on what I was replying to, probably not that relevant anymore, sorry about it

[ramseysounds]

Here we go again