Modelers and Aliasing

[yeky83]

I was assuming for guitar y'all didn't want anything beyond 10k? So there was an opportunity to filter a bit more

The Kemper did brickwall at 11kHz for some time (not anymore) and no one seemed to notice until a certain engineer pointed it out, so you might have a point

 

[DLC86]

Anyway, all this talk about filtering the signal either before or after doesn't make sense to me. Aliasing is produced by non-linearities (aka distorion in this case) so the only way to attenuate it is producing the non-linearity at a higher sample rate, filtering out what's beyond 20kHz and then downsampling back, there's no other way around.

 

[imperator]

For what it's worth, I put a tone a 11k into nam and the aliased peaks were like 90 dB down for the savage model.

 

[FractalAudio]

The input to an amp can easily reach ultrasonic frequencies. The initial pick attack has copious high frequency content.

Many distortion pedals have an output spectrum that reaches far beyond 20kHz.

If you limit the frequency to 10kHz it will sound dull.

None of that really matters though. What matters is the fundamental concept of aliasing which some here don't seem to grasp. Any time you have a nonlinear transfer function you create distortion. In guitar gear the transfer function is usually some sort of clipping behavior. At the limit this turns a sine wave into a square wave. A square wave has harmonics that extend well into the ultrasonic range.

If you don't oversample enough those harmonics alias into the audible spectrum. No amount of hand-waving changes that. The correct thing to do is to increase the sample rate and then downsample after all the processing is done.

 

[pipelineaudio]

I was assuming for guitar y'all didn't want anything beyond 10k? So there was an opportunity to filter a bit more

Its not the input signal bandwidth that is the concern here...I can show aliasing by messing with a 20hz sine wav....its actually one of the bugaboos for making compressor plugins for bass guitar where you really hear it

 

[TubeStack]

"Aliasing.... aliasing.... I hear aliasing..... aliasing over my head."

This keeps running in my head now, with these lyrics…

 

[DLC86]

And for those that still haven't seen this...

 

[pipelineaudio]

I'm trying to find any of the threads where he said it, but I think Dan Lavry said something like even 128x oversampling wouldn't be enough for making compressors audibly unaffected by aliasing. I know for sure ABX testing proved that wrong with at least human hearing or tolerance, BUT he was able to show it on pathological files that even extreme oversampling could still at least show the stuff on a graph.

On the video I did, posted earlier on the thread, I think I showed 64x or 128x oversampling doing almost nothing to fix the problem, but again, its not any sort of signal you'd ever see in recording or mixing a band.

For you amp modeling experts: how high does oversampling need to be before its not really noticed in say an ABX test?

I personally often mistake IMD for aliasing

 

[norminal]

RE: Aliasing

Serious question:

Is it audible in a mix with distorted rhythm guitars that are band-passed between, say 100-8,000 Hz?

 

[DLC86]

For you amp modeling experts: how high does oversampling need to be before its not really noticed in say an ABX test?

I guess it depends firstly on the amount of gain used, a totally clean amp could probably run with no oversampling at all and you won't notice it, a high gain amp otoh probably needs at least 8x or 16x. I think the aliasing relative level is a better parameter to tell.

PS: and I suspect it might not be that simple considering all the complex stuff that happens inside an amp

 

[yeky83]

Maybe most of the leftover in these null tests is aliasing noise?

 

[pipelineaudio]

When you say, hold the b string at the 10th fret and the g at the 12th fret, and pick them at the same time and bend only the g, is that weird ghost note I'm hearing IMD or just aliasing? It gets lower and lower in volume as the sample rate goes up, but with a decent amount of distortion, you can certainly hear it in a mix even at 2x oversampling

 

[norminal]

Maybe most of the leftover in these null tests is aliasing noise?

Sloppy seconds.

 

[James Freeman]

Maybe most of the leftover in these null tests is aliasing noise?

Try that with a sine tone.

 

[signalpath]

Probably not related to aliasing since the Stomp is aliasing all the same into a cab (which to be fair could be masking the effects of aliasing I suppose...), sounds to me like your ear-brain just doesn't like playing guitar to the sound of mic'ed cabs.

That makes sense but the reason I raise the effect as being potentially related to aliasing is that if what you suggest is true it would be exactly the same for all modellers running into an IR. And it isn’t. As I said the onset of the effect has been reduced in the HX family since the over sampling update and the FM3 is better still. And if I really wanted to chase this down I’d get an Axe-FX III and a FM9 and see what happens. But my pocket categorically refuses to countenance that proposition

If the effect is being triggered by aliasing in the high frequencies, which by my understanding is where a lot the aliasing happens, then as you say maybe the guitar cab is masking that.

 

[Bob Zaod]

Oh and here I was thinking this was a thread about the show Alias.

 

[yeky83]

Some idiot made this one a few years ago

On the video I did, posted earlier on the thread, I think I showed 64x or 128x oversampling doing almost nothing to fix the problem, but again, its not any sort of signal you'd ever see in recording or mixing a band.

Either the plugin's oversampling feature's not working or your setup is wrong, but sounds like the oversampling change isn't doing anything in the video. When you flip through x2 to x64 at ~4:30, they don't sound or appear any different. Probably want to try again.

 

[pipelineaudio]

Either the plugin's oversampling feature's not working or your setup is wrong, but sounds like the oversampling change isn't doing anything in the video. When you flip and compare x2 to x64 at ~4:30, they don't sound any different. Probably want to try again.

That was the point. At least at the time, for really high gain (or extreme compressor settings), like Lavry said, even super high oversampling rates would still be audible

 

[yeky83]

That was the point. At least at the time, for really high gain (or extreme compressor settings), like Lavry said, even super high oversampling rates would still be audible

No, they should sound markedly different. See the video @DLC86 posted above: https://thegearforum.com/threads/modelers-and-aliasing.1815/page-5#post-59854

 

[DLC86]

When you say, hold the b string at the 10th fret and the g at the 12th fret, and pick them at the same time and bend only the g, is that weird ghost note I'm hearing IMD or just aliasing? It gets lower and lower in volume as the sample rate goes up, but with a decent amount of distortion, you can certainly hear it in a mix even at 2x oversampling

I think that's just IMD plus beat frequency (the latter happens even on acoustic instruments cuz it's just interference between two nearing frequencies)