Analog filter automating resonance produces glitches


I found the analog filters produce glitches when the resonance is automated. They are very quiet glitches, in normal operation the peaks are barely visible in the spectrum view. But when using advanced techniques such as stacking filters, and/or boosting the crap out of the results with drive or distortion devices, they can become more and more problematic.

There seem to be 2 kinds of glitches going on when automating resonance. The first is a piercing shrill noise at 1/8 the frequency the filter is running at. So global sample rate and oversampling the filter affects the frequency of the noise. A filter with renoise running at 48khz with no oversampling produces a peak at 6khz, with 2x oversampling 12 khz, with 4x oversampling a theoretical 24 khz. I verified this by running renoise at 44.1 khz, the peaks correspond.

The other glitch is less problematic and could only be audible reproduced by me by feeding dc into the filter. It sounds more like a train of popping when driven, in the frequency that the resonance is automated. So a lazy midi controller controlling reso will make the filter sound a bit like zipper noise rubbing over rubber, and automating with graphical automation makes it a tonal noise, probably in the frequency that the automation data is passed to the filter.

The first noise can obviously be worked around by putting the relevant filters into 4x or 8x oversampling mode, at cost of much higher cpu usage. The second noise can only be worked around by NOT automating resonance, well, at least it seems not as problematic as the first anyways.

I created a little bug showcase instrument where the effect is audible, also to show that massively boosting a filter’s output isn’t esoteric bla bla, but a normal sound design technique capable of yielding very dope sounds. It is a very simplified version of one of my instruments suffering from these glitches, I stripped off almost everything that has nothing to do with the glitches, added controls to further explore them, but still tried to make it sound not overly boring.

Steps to reproduce:

  1. make sure you run renoise at either 44.1 or 48 khz, load the attached instrument

  2. sequence notes and play them in a loop (like around the c-3 to c-4 range) or constantly jazz on a keyboard to produce output

  3. automate the instruments first macro (“Glitch Me”) over the full range, either with graphical automation in line or curve mode. Create periods where this macro is in constant moderately slow movement, and periods where it is kept totally constant for comparison.

(Edit: forgot to mention, using a midi controller with low time resolution or point mode automation with moderately high res gives better impression of the first glitch, as the second will turn from whine to rubber rubbing, leaving more space for the shrill noise)

  1. you should hear a scratching shrill noise at 6000 or 5512.5 hz, drowned a bit in the instrument but making it sound unpleasant, while moving the macro. This is effect of the first glitch.

  2. Move the “Automate Reso On Off” macro from zero to 100, this will turn off automation of the resonances, the glitch seems gone, but the bass won’t scream anymore and not sound as soft in the lower cutoffs, making it less defined.

  3. turn back the On Off macro from step 5 to below 50, and turn down the “Bass” macro from 50 to zero, and the “DC” macro a bit, to hear the glitches without the bass sound. You will now hear the second glitch, and the first in clearer fashion.

  4. reload the instrument, and set oversampling for the two filters “Sorgenkind1” and “Sorgenkind2” to 2x - the frequency of the first glitch will double. Turn both to 4x or 8x, and the glitch isn’t audible any more. If you repeat the DC step 6 without the bass in this mode, you will only hear the zipper noise, not the shrill one anymore.

As this is very important for me - many of my sounds depend on automating resonances - are there any prospects of getting at least the first glitch fixed at some time? I have many instruments that suffer, most only a tiny bit and not as extreme as this example, but it is still disturbing to have such dissonant noises in my sounds that might make the listener feel incomfortable without him knowing why.

Thank you for attention

yep , I noticed this too , posted it during beta .

Maybe we are the only ones blessed with an exceptional good hearing .

The worst about it is: that 5.5/6 or 10/12 khz whine is just right inside that dentists torture tool frequency, that everyone recognizes subconsciously & painfully. I found that instruments where even I don’t immediately notice, because it is too quiet or drowned in other sounds, there is an unpleasant pressure in perception of the sound, that gets lifted massively for me when I oversample all reso automated filters to 8x. But oversampling them all is no real solution, it really eats the cpu for breakfast.

On the positive side: we now know about some renoise-native glitches to sample and have fun with. Cool robo-batcalls and rubber farts for the masses. haha, just kidding. The filters are great otherwise, I really love them and use gazillions of them. Humble, effective, and have their own character. please fix them!