Simple device for oversampling

[jbone1313]

I already did a search, but I have not found anything.

Is there a simple M4L device(s) which can add oversampling to any plugin? Kind of like what DDMF Metaplugin does.

Is this even possible in M4L? I.e., does M4L allow oversampling?

If it is possible, then I imagine there would need to be two devices, so that the plugins we want to oversample would sit between them. Or, there would need to be some way to tell the device(s) the plugins to which oversampling should be applied.

[pottering]

Not aware of any such M4L device either.

Is this even possible in M4L? I.e., does M4L allow oversampling?

May be possible, as Max allows oversampling in "poly" and inside Patchers/BPatchers (which are all kinda like "sub-patches" in practice).

But benefit may be small (good plugins either already sound good in 44.1 or have their own oversampling options, lame plugins with less aliasing are still lame), while the CPU hit disadvantage is immediate and very noticeable (2x oversampling doubles CPU, 4x quadruples, and so on).

And if users end up using oversampling with most of their plugins (because they can't really tell the difference to know which plugin should be oversampled, but are told oversampled is better), the CPU hit ends up nearly the same as just rising the sample-rate in the Audio Prefs (which is much easier).

there would need to be two devices, so that the plugins we want to oversample would sit between them

Can't be done that way, has to be just like Metaplugin does, loading the plugin inside the sub-host, inside the M4L device in M4L's case.

IMO, not worth the time trying to fix those issues with VSTs from outside, easier to buy trustworthy plugins from devs like Cytomic, and just ditch plugins whose sound quality you don't trust. Or run Live in higher sample-rates.

Too much work if it is just for a few plugins you use, too much CPU (compared with just running DAW in higher SR) if it is most plugins you use.

And oversampling single plugins implies a downsampling step after the oversampled processing, downsampling has it is own problems, it is not transparent (maybe the upsampling is not transparent either), so even the premise of "oversampling==better quality" is flawed, depending on the audio you could have no benefit from oversampling yet have downsampling artifacts (not sure really, but it may be similar to using samples of different sample-rates in the same project).

[jbone1313]

Thank you for the reply. Just FYI, the reason I am curious about this is I think it would be interesting to try to run some Airwindows plugins oversampled.

Also, in case you have not yet seen this video from Dan Worrall, he makes a compelling case that oversampling is much better than running a whole project at a higher sample rate. It is worth a watch.

https://www.youtube.com/watch?v=-jCwIsT ... x=7&t=721s

[pottering]

Yeah, I see what you mean in that video, I changed my mind, thanks for the video.

96 kHz is about x2.18 "oversampling", that video shows 2x may not be enough to push aliasing outside the hearing range.

It shows a High Quality option in the Saturn plugin, seems to be up to 32x, its manual says 8x is a good compromise.

8x would be 352 kHz (!) if done on the DAW.

So per plugin options would be more usable.

It is possible in M4L, it is just that loading VST is a bit of a PITA in Max (I mean mostly preset recall/automation, just loading is not that hard), even harder if you want a nice M4L device that looks like a Live device (instead of a Max patch with spaghetti cables).

Though those Airwindows plugins are simple, should be much easier with them.

From what I read, in Max you can use options in "poly~", or maybe make a sub-patcher with different sample-rate (not sure that works).

https://cycling74.com/forums/oversampling-with-poly
https://cycling74.com/forums/real-sampl ... -of-buffer~
https://cycling74.com/forums/implementi ... aper-patch
https://docs.cycling74.com/max8/tutoria ... ing%20rate
https://cycling74.com/forums/loadbang-sample-rate
https://docs.cycling74.com/max8/vignett ... ges_to_dsp

[jbone1313]

Thanks much for the reply. You’ve given me much to think about.

Cheers