List of Oversampled Plugins?

[Retro Audio Enthusiast]

Is there a list of over-sampled plugins ? Or, if anyone knows which companies use it?
I know Klanghelm's plugins all use oversampling.

Thanks in advance.

[Pinknoise]

Melda, Kazrog, Hornet...

[future fields]

Fabfilter Saturn
Cytomic The Glue
Vladg Limiter no6

^ these have switchable on/off oversampling

[brainwreck]

Vladg Molot

[Tinus]

jsCompShaper

Kotelnikov and Nova by Tokyo Dawn Labs have oversampling too I believe. Not sure though.

[Softsynth]

Originally Posted by Tinus

jsCompShaper

Kotelnikov and Nova by Tokyo Dawn Labs have oversampling too I believe. Not sure though.

FWIW (from the manual):

NOVA GE offers five different Quality modes. These primarily affect anti-aliasing:
 Eco: Economic mode, with an internal bandwidth ~100kHz
 Precise: Default quality, with an internal bandwidth ~200kHz
 Insane: Very high quality, with an internal bandwidth ~400kHz
 Precise+: Precise quality, with internal nonlinearity.
 Insane+: Insane quality, with internal nonlinearity.
Set quality for all instances sets the currently selected quality mode to all

[sai'ke]

I have oversampling in my nonlinear stuffs:
Squashman (multiband saturator), Tight compressor, Transience, MS-20, Wahriffic
Reapack link: https://raw.githubusercontent.com/Jo...ster/index.xml

Filther also has oversampling, but I've been meaning to rework the oversampling with a steeper brick-wall filter (the one already in Squashman now).

Most are pretty CPU hungry when you oversample though.

Nitsuj's ReEq also has oversampling, although not to prevent aliasing (since they're linear filters), but to prevent filter cramping.

[ernzo]

Izotope Ozone, TB GEQ12, Elysia Alpha Master Comp..

There's many more but I forgot :P

[brainwreck]

On the amp sim front, I'm not sure which ones have oversampling. I know that Amplitube does, but it still aliases like crazy. I would think that all of them use oversampling, but with such a high amount of distortion, I think it can't work magic.

[Judders]

Originally Posted by brainwreck

On the amp sim front, I'm not sure which ones have oversampling. I know that Amplitube does, but it still aliases like crazy. I would think that all of them use oversampling, but with such a high amount of distortion, I think it can't work magic.

You can't have oversampling without introducing latency, which is obviously bad for playing guitar.

[Pinknoise]

Originally Posted by brainwreck

On the amp sim front, I'm not sure which ones have oversampling. I know that Amplitube does, but it still aliases like crazy. I would think that all of them use oversampling, but with such a high amount of distortion, I think it can't work magic.

KAzrog has 5 Quality settings and Kuassa has normal and HQ.

[brainwreck]

I think maybe a false assumption is being made here, that oversampling necessarily elliminates aliasing. It doesn't. It reduces it. And so then, how much oversampling (and filtering) is required so that aliasing is never an issue? Or, can aliasing practically be elliminated at all?

[Judders]

Originally Posted by brainwreck

I think maybe a false assumption is being made here, that oversampling necessarily elliminates aliasing. It doesn't. It reduces it. And so then, how much oversampling (and filtering) is required so that aliasing is never an issue? Or, can aliasing practically be elliminated at all?

The audible effect of aliasing is contextual. In many situations it will be masked and so effectively not an issue at all.

[brainwreck]

Originally Posted by Judders

The audible effect of aliasing is contextual. In many situations it will be masked and so effectively not an issue at all.

I guess if you have some processing that generates buckets of harmonics, such as an amp sim for the high gain 'djent' stuff, the aliased frequencies could get masked in all that, at least without some direct comparison to the same sort of processing that is alias free (analog). But for anything else, I don't see how it could be practically masked.

[karbomusic]

Wouldn't you just find one that does that, then raise the sample rate as a test to confirm that's what it is?

[Judders]

Originally Posted by brainwreck

I guess if you have some processing that generates buckets of harmonics, such as an amp sim for the high gain 'djent' stuff, the aliased frequencies could get masked in all that, at least without some direct comparison to the same sort of processing that is alias free (analog). But for anything else, I don't see how it could be practically masked.

Because, unless you are playing sine waves through a synth, your signal will be broadband. High gain use of amp sims is more likely to produce audible aliasing, not less.

[brainwreck]

Originally Posted by karbomusic

Wouldn't you just find one that does that, then raise the sample rate as a test to confirm that's what it is?

I wish it were that easy. In some cases, so much aliasing is happening that that running at say 192k hz just isn't enough to matter. And some of those plugins have anti-aliasing that also doesn't seem to be doing anything audible, even at highest oversampling settings. I'm not at my machine at the moment, but maybe Reaper's dummy driver goes to high enough sampling rates to matter for some testing.

[karbomusic]

I think it should be something audible/demonstrable enough or similar for it to be a real problem in context of whatever it is. I'm not saying it isn't one, just seems like the first step is to confirm what one hears (that they don't like) is actually aliasing. Secondly, I'd think 192 would make it roughly 4 times better and would help confirm.

[brainwreck]

Originally Posted by Judders

Because, unless you are playing sine waves through a synth, your signal will be broadband. High gain use of amp sims is more likely to produce audible aliasing, not less.

Sine waves as test signals doesn't really have anything to do with it. A real signal is going to have a ton more aliasing than a sine. A sine is really best case (least aliasing). And I disagree on high gain having more audible aliasing. With a semi-clean signal, there are far fewer harmonics to hide the aliased frequencies, which is likely why amp sims tend to sound so bad at lower gain settings, and turning up the gain hides some of the ugly, by quite a bit actually. It's an issue of contrast. I think this has played a major role in more guitar players using high gain with amp sims.

[brainwreck]

Originally Posted by karbomusic

I think it should be something audible/demonstrable enough or similar for it to be a real problem in context of whatever it is. I'm not saying it isn't one, just seems like the first step is to confirm what one hears (that they don't like) is actually aliasing. Secondly, I'd think 192 would make it roughly 4 times better and would help confirm.

It definitely is audible, but the problem is in getting an alias free signal from the same processor for direct comparison.

[karbomusic]

Why wouldn't it be less audible oversampled? I don't mean gone, I mean less. At minimum the foldback/reflected frequency would move higher, and that could be measured and/or heard, no?

On a side note, the infinite or near infinite harmonic series (which you get with a clipped-to-square wave and is worst case scenario) does have a very steady rolloff so I'd imagine this would work in our favor with the above at some point.

[brainwreck]

Originally Posted by karbomusic

Why wouldn't it be less audible oversampled? I don't mean gone, I mean less. At minimum the foldback/reflected frequency would move higher, and that could be measured and/or heard, no?

On a side note, the infinite or near infinite harmonic series (which you get with a clipped-to-square wave and is worst case scenario) does have a very steady rolloff so I'd imagine this would work in our favor with the above at some point.

I think it is a matter of how much aliased content is in the output and what the threshold is for 'better'. Silly analogy: If your head is in a beehive of a thousand angry bees and half of them are taken away, is it going to be 'better', or might you rather be dealing with say, a few bees before things are 'better' for you? What is the threshold? It likely isn't a linear issue. In analog audio, we can be turned off by very small amounts of hard clipping in contrast to an otherwise clean signal, for example.

[karbomusic]

Originally Posted by brainwreck

I think it is a matter of how much aliased content is in the output and what the threshold is for 'better'. Silly analogy: If your head is in a beehive of a thousand angry bees and half of them are taken away, is it going to be 'better', or might you rather be dealing with say, a few bees before things are 'better' for you? What is the threshold? It probably isn't a linear issue.

Yes, I think it would make it demonstrably better FWIW. I don't mean whether the person likes the SIM better but being able to demonstrably understand that it's the aliasing and, how much it does change. If the gain is the same, I expect it to be fairly linear as far as the harmonic falloff is concerned; that's easy to check though.

[brainwreck]

I tried rendering using Reaper's dummy driver. I guess it doesn't actually allow for that. I got blank renders. I don't know how useful it would be anyway, given that no filtering can be applied inbetween clipping stages before downsampling. Maybe there are some plugins with single clipping stages that would be useful for testing though.

What I don't know here is if oversampling plugins tend to have limits on the frequency range. Say that I record some audio at 192k, with a plugin that oversamples x8, is the plugin actually going to oversample to 1.536Mhz?

Also, it might be the case that we don't want any filtering going on, so that we can experiment with various filters. There isn't going to be any way to disable internal anti-aliasing filtering.

So then, maybe it would be useful to try and knock up a clipping plugin in js for testing purposes. Although, audio filtering is currently out of my range of abilities in js. But I'm reading up on it.

Maybe someone else has some immediately practical ideas on how to go about this.

[Nip]

I used Metaplugin with 2x oversampling some years ago - before Sonar got that as standard. So you run selected plugins loaded at double sample rate compared to project.

You can load an instrument directly from disk - and put oversampling on. In my case I had to save and reload project for that to follow through.

This is different than running a full project at 96k - since you have to do both upsampling and downsampling within the project. This shows first on a 96k project when final render to 44k.

Depending on content it is clearly audible - like I did a test on strings which is very rich on harmonics. So doing chords it also did a bit less harmonic beating - more supressed.

Not so sure how amp sims differ - if odd harmonics in content to a large extent it might not be as obvious. It is subtle so not night and day either case.

But if running a bunch of instruments with oversampling - it also sums up on each instance a little better - makes a full mix clearly better. Some has described it as more open sounding.

The idea is the same on using high end preamps, like $2000 ones instead of $1000 ones - audible but maybe not night and day in one track - but total is summed up being much nicer sounding mix. Probably why million dollar studios spend that extra money on ones that a home studio would not.

[karbomusic]

@Brainwreck, dummy audio seems to render here if that helps.

Regarding odd harmonics, it should make no practical audible difference (I can easily demonstrate with a signal to some extent), they aren't special compared to even in the way most think they are. Regarding preamps, I have such nice preamps, see my sig; they would all be uncorrelated tracks and it would be insignificant anyway unless each preamp was pushed into distortion.

The only credible difference I remember with odd vs even harmonics is when the distortion is as loud/louder than the original signal, where the 2nd harmonic is quite loud (due to filtering of higher harmonics so that it becomes audible), this is heard as an octave along side the 1st harmonic and is what makes say a fuzz pedal (around 12th fret, neck pickup), and some amps sound cool but that's about as much distortion as a circuit can create aka a square wave so it's magnitudes beyond what normal clean preamp audio would encounter.