LinuxMusicians

Frank Carvalho wrote:But that leaves the question, why don't they play nice at 96kHz?

That's cause of the IIr filters. They become unstable on higher sample rates.
What I could do is implement a fixed rate resampler which take care that the plug run internal at 48kHz regardless which sample rate you use.

I am glad that you know the reason. But I think it is a problem, as any serious user would want to run the simulations at higher sample rates. From my experience with the RPi, I can say, that 44.1kHz is just too low quality for serious use. The sound gets harsh and feels limited. 48kHz is marginally better.
As a quick fix, resampling to 48kHz is obviously better than the current behaviour. Best thing, though, would be to support sample rates all the way up to 192kHz.

/Frank

Frank Carvalho wrote: I can say, that 44.1kHz is just too low quality for serious use. The sound gets harsh and feels limited. 48kHz is marginally better.
As a quick fix, resampling to 48kHz is obviously better than the current behaviour. Best thing, though, would be to support sample rates all the way up to 192kHz.

I'm doing my first experiments of running the linear IIr filters at a fixed rate of 48kHz and do the non-linear processing oversampled (4x, 6x).
This approve themes very promising to me, as the used CPU power increase very, very little. By doing only the real non-linear part oversampled 6x I get here a CPU increase of 1% on a Quad core Intel Core2 with 2360 MHz.

Unfortunately creating IIR filters to work on any sample freq is like impossible. Only first order IIR filters "may" work.

Imagine, it's like using IR-files for convolution, they comes with a sample-freq, which you may need to resample before use. ( True, usually your convolver engine will do it on the fly for you).

One plug which already include a fixed rate resampler is the GxSVT.lv2 plug. (the Ampeg Bass Preamp), may you like to check if you find any issues with that one?
If not, I would like to implement that approve to all GxPlugins.

regards
hermann

I could try it out at 96kHz, and see how it behaves.
Would it perhaps be a possibility, to build the filters to work at a few selected frequencies, say, 48kHz, 96kHz and 192kHz, which are the most common ones?
Alternatively, why not build the filters with a high internal sample frequency like 96kHz or 192kHz, and use upsampling/downsampling to work in 44.1 and 48 environments? Just some suggestions. I have no idea how difficult this is to achieve.

/Frank

Frank Carvalho wrote: Would it perhaps be a possibility, to build the filters to work at a few selected frequencies, say, 48kHz, 96kHz and 192kHz, which are the most common ones?

Well, may be possible, just I ain't own a sound-card which support those sample rates, so I ain't could test that.

Frank Carvalho wrote: Alternatively, why not build the filters with a high internal sample frequency like 96kHz or 192kHz, and use upsampling/downsampling to work in 44.1 and 48 environments? Just some suggestions. I have no idea how difficult this is to achieve.

/Frank

Drawback of this will be that the CPU usage increase when we upsample for the IIR filters for no real benefit.
Higher sample rates been mostly useful for non-linear processing. On linear processes, you wont be able to notice a down/up/sampling in the result.

So I've pushed the first plug which run internal at a fixed sample rate of 48kHz and oversample for the non-linear part about x4.
From my experiences this is a very useful combination.
https://github.com/brummer10/GxTubeDistortion.lv2

I'll check it out at 96kHz. (Been looking for a tube distortion for a while BTW, to emulate my Vox BigBen Overdrive).
My main plugins are the SuperFuzz and GxVMK2, and they both act unstable at 96kHz. Would it be a lot of work to redo these plugins?

/Frank

Frank Carvalho wrote:My main plugins are the SuperFuzz and GxVMK2, and they both act unstable at 96kHz. Would it be a lot of work to redo these plugins?

No, it wont.
I'll look in it tomorrow.

Thanks again Mr Tramp, Guitarix is just too much fun \m/

asbak wrote:Thanks again Mr Tramp, Guitarix is just too much fun \m/

Thanks asbak!

Feedback, as special when it praise my work, is highly appreciated here.

Frank Carvalho wrote:I'll check it out at 96kHz. (Been looking for a tube distortion for a while BTW, to emulate my Vox BigBen Overdrive).

Well, this circuit use 12au7 tubes, so the distortion is very, very mild. Indeed more like a overdrive.
It sounds beautifully to me when using before GxAmplifier.

Btw. do you know a schematic file online of the Vox BigBen Overdrive?

I found the schematics here:
https://www.eserviceinfo.com/download_s ... rlid=44677
Also based on a 12AU7 it seems.

I tested the GxTubeDistortion. Very nice sound, indeed!
It works perfectly at 96kHz - no trouble at all. Funny thing, the entire setup has more headroom at 96kHz, which I find really nice. Probably some 3-6db louder with the exact same settings. I like that a lot.

Frank Carvalho wrote:I tested the GxTubeDistortion. Very nice sound, indeed!
It works perfectly at 96kHz - no trouble at all.

Thanks for testing, good to hear it works.

So I've implement this architecture as well for the GxSuperFuzz and the GxVMK2.
You'll get the updates when you do a fresh checkout on the GxPlugins git repository.

Very cool. Thanks! These are great plugins.
I'll check them out.

tramp wrote:

Frank Carvalho wrote:I tested the GxTubeDistortion. Very nice sound, indeed!
It works perfectly at 96kHz - no trouble at all.

Thanks for testing, good to hear it works.

So I've implement this architecture as well for the GxSuperFuzz and the GxVMK2.
You'll get the updates when you do a fresh checkout on the GxPlugins git repository.

Bud, are you still making binaries of these plugins? I did a fresh install of arch about a month-and-a-half ago and now the binaries are showing up with proper GUI in Mixbus and ardour.

What pedal board software is that?

tramp wrote:Actual this combo give me the kick I'm after: