@AFKAEjay(retired) - this is what I get when I post sincere and polite responses to people. So... please tell me why I should bother; coz you are extremely fond of calling me out by stealth, yet you never call these kinds of people out. Why tone police me, and not other people?
As to the claim that the evidence is to the contrary... doesn't even warrant a response. It is just stupid, and the entire planet has just lost half an IQ point because of it.
I've got 18 years of gigging under my belt too. But you'll just come back with another stupid response to that, is my guess. You're an extremely bad faith person.
Aliasing is not a per-rig phenomenon. It is a product of the entire DSP system. It is genuinely hilarious that you think you can avoid it by just avoiding certain rigs. You reveal your ignorance once again.
Well, firstly, it isn't off subject. It is precisely what we're talking about.
Aliasing happens when you apply non-linear processes to a signal, that create overtones or harmonics. These harmonics extend out from the fundamental, up through the frequency range. When they hit Nyquist, if they're not dealt with, they will wrap around and cause unpleasant chipmunk or squirrel artefacts. In the case of the Kemper - as the video I linked earlier shows - they're extremely audible.
How do you handle aliasing? You filter. How do you filter? There are a range of techniques. One of them is to apply a FIR filter just below the Nyquist frequency, so that frequencies above bounce back and get filtered before they have a chance to be audible.
Another way is to perform an FFT on the signal, attenuate or zero out bins that fall above a certain threshold, and then inverse FFT it back. This method, while computationally more expensive, can offer precise control over the frequency domain and is often used in convolution-style processes or spectral shaping tools.
But most importantly — and here's where many profiling and modeling systems fall short — the best way to mitigate aliasing is to
oversample the non-linear process
before it generates the alias-prone content. You run your distortion (or other non-linear effect) at, say, 4x or 8x the sample rate, then filter and downsample it back. This dramatically reduces foldback aliasing.
It is way more involved than simply saying "run at 48kHz and everything will be fine".
So to answer your question:
=> IR block - very low risk of aliasing being introduced here.
=> Delay Blocks - moderate risk of aliasing here, depending on how the delay line interpolation works and the non-linearities added in the circuit.
=> Reverb Block - same as the delay block; there's a possibility you could introduce aliasing here if you don't support fractional delay times within the feedback paths used to create the reverb.
=> EQ- very low risk of aliasing being introduced here.
=> MultiBand Comp - There could be some risk here, depending on non-linearities and the design of the crossover filters.
=> Flanger / Phaser - Same sort of thing as the delay and reverb blocks; if you're modulating delay lines, then you could introduce aliasing.
Using band-limited interpolation techniques and oversampling within the blocks where delay lines are modulated, is how you would address any potential sources of aliasing.
