Oversampling: When/where should you use it?

abaga129

The Apprentice
Earlier I was browsing different free vsts and came across an oversampling/anti-aliasing plugin. I realized that I knew almost nothing about oversampling and decided to do some research. After reading two very informative articles I feel that I have a very good basic understanding of aliasing and oversampling but I am still curious about when to use oversampling.

The plugin that sparked my research was just simply a little box with options for x2 x4 x8. Would you use this as the first insert effect or would you place it at a certain step in the chain. Also what instruments usually benefit from oversampling?

My final question is are you better off to just record at 96k and not have to worry about any of this?

I hope that other people can benefit from this thread as well because it seems to be a less discussed subject.

Thanks guys!
 
Set your oversampling to max for every plugin you are running if your cpu can handle it. It's gonna be less likely to have artifacts this way, but it's not that lower settings are going to damage your audio, at least in my experience, and it doesn't matter where you place it in the efects chain. If you are only doing audio in stereo I wouldn't worry about going above 44.1 khz

The plugin that sparked my research was just simply a little box with options for x2 x4 x8. Would you use this as the first insert effect or would you place it at a certain step in the chain.

EDIT: What little box? Care to post a screenshot?
 
If you're running at 88.2k or 96k sample-rate, you don't need oversampling. You can turn it off. All you're doing is taxing your CPU for no additional benefit. On my i7 setup, I run at 44.1khz and I generally go to x2 - because that gives the plugin an internal sample-rate of 88.2khz. But you need to evaluate it with your ears, because not all plugins are made alike. Depending on their downsampling mechanism at the end of their DSP chain, it might actually sound WORSE than leaving it off.
 
oversampling is generally better for things involving distortion. You'll notice more of a difference on those than anything else. A lot depends on the plugin at hand and the way its designed, regarding what (if any) improvements you'll get. There's way too many variables to make blanket statements about it.
 
here we go again...

Nah, no need for that. If something sounds better, why settle for something inferior?

With modern CPUs, you can bombard your computer with hundreds of tracks and plugins while retaining super short latency even when working at 192khz. I see no reason to work at 44.1 when 96khz yields better results.

E: Of course, in plugins, oversampling helps with the unwanted artifacts, when working in 44.1khz. But that doesn't help with the information you already lost and artifacts you gained when you track'd it 44.1.
 
You do understand nyquist theory, right?

Only the basic consept. I don't have the technical nor mathematical know-how to go deeper than that.. Nor do I really care for..

I base this mostly on my gut, ears and info that I've gathered from different people. Be they algorithm designers or audio engineers. I don't know if there's less shit or more good stuff, or both in 96Khz, but things sound different. And for the better..

I suck at A/B's and critical comparisons. I just like to fuck shit up 'till it makes me headbang and play airdrums.

And as much as I hate name-dropping type argumentation, there are hundreds and thousands of seasoned pro's who agree with this. Dudes who'll forget more about this shit then I'll ever learn. My logic says that if theory says that there can be no measurable / audible differences, yet people hear these.. Then we need to refine our ways of measurement and thinking.

E: Or I might be full of shit. That has happened before too.
 
:lol:

Well I think people pay attention way too much to numbers and not their ears. But according to the theory, no information that we can actually perceive is lost when recording at 44.1khz. That's enough for me to not care about cramming hard-drives full of extra data.

YMMV.
 
I think this article sums things up decently. It doesn't go in depth on everything, but it's easy to find out more about it yourself.

I think that if convertors are up to snuff, there should be no difference between the regular and higher samplerates. As said in the article, if you really hear a difference, it has to be either because your converters have been designed to run at the higher samplerate, or because you are hearing inter-modulation distortion. Not even to mention that many recording-tools aren't even made to capture frequencies above our hearing-limits.
So it could very well be that you hear a difference between regular and high samplerates. But it may only be like that on your own convertor, and not universal truth.

Plugins are different though. Like Machinated said, oversampling can be especially useful on plugins that introduce some form of distortion/saturation. This has to do with them adding harmonics. this guy explains that pretty well.
 
If I remember correctly you use VCC right? Render an entire mix with 0x oversampling and then with 4x. I think there is quite a big difference there :)

No....I only ever bought 1 plugin and it's the $20 SSD Sampler....

But I imagine VCC is something you'd throw on your master bus?

I've only ever "tried" oversampling with amp sims and TSE808 plugins on an individual track. Also - never messed with dithering. I see there is an option to enable it in Reaper when rendering down....maybe I need to start experimenting.
 
No....I only ever bought 1 plugin and it's the $20 SSD Sampler....

But I imagine VCC is something you'd throw on your master bus?

I've only ever "tried" oversampling with amp sims and TSE808 plugins on an individual track. Also - never messed with dithering. I see there is an option to enable it in Reaper when rendering down....maybe I need to start experimenting.

Heh, must have been someone else I had in mind then :D

Yes, VCC is something you throw on the masterbus and all over the mix typically, so that's why I think it's a good example of what oversampling can do. It's still not "omg version A sounds like rocks and version B like diamonds", and some people even prefer the 0x oversampling sound. It's just slightly different. Over time I've settled on 4x oversampling with it, because I preferred it in my own blind tests almost universally.

So yes, experiment! But if you don't notice a difference, it's all good!
 
I think this article sums things up decently. It doesn't go in depth on everything, but it's easy to find out more about it yourself.

I think that if convertors are up to snuff, there should be no difference between the regular and higher samplerates. As said in the article, if you really hear a difference, it has to be either because your converters have been designed to run at the higher samplerate, or because you are hearing inter-modulation distortion. Not even to mention that many recording-tools aren't even made to capture frequencies above our hearing-limits.
So it could very well be that you hear a difference between regular and high samplerates. But it may only be like that on your own convertor, and not universal truth.

Plugins are different though. Like Machinated said, oversampling can be especially useful on plugins that introduce some form of distortion/saturation. This has to do with them adding harmonics. this guy explains that pretty well.

Thanks Nimvi for the read. One can always learn something new.

But something crossed my mind, that wasn't explained on either article.. or atleast I think it wasn't.

The consensus seems to be that for plugins = oversampling good because harmonics and aliasing. Recording at high sample rates = Grab your tin foil hats, folks!

Overtones and harmonics don't live in a vacuum inside saturation plugins. Excluding sine waves, sounds are their overtones and harmonics. That's what makes them sounds.

We record sound sources in rooms, with gear and everything we record, contains that type of information. We pummel the signal trough pre amps, compressors, eq's, saturators, vca's, AD/DA's, tubes, cords, circuit boards etc. They all add different types of distortions, harmonics, modulations and god knows what.

Why would this NOT cause aliasing? If we worry about the harmonics and overtones that our plugins introduce, why shouldn't we worry about the harmonics and overtones that are inherit to the signals we record? Or the harmonics and distortions that our gear introduces?
 
Correct me if I'm wrong but we do worry a lot about the harmonics and distortion that occur in our hardware and whether or not they are desireable harmonics is what sets the bad gear from the great. And aliasing doesn't occur in analog hardware since the sample rate is technically infinite because its an electrical signal that is being dealt with versus samples in a daw.