Questions about quantizers, rhythm, and probability

I have been thinking about semi generative pitch modulations lately, both in and out of the ER-301. Of course quantizers are key when it comes to this but I often find it difficult to get results I like. One quirk of quantizers I noticed (both in my intellijel ustep and the ER-301 quantizer) is the way they assign incoming pitches to the chosen quantized values. You can imagine each note chosen has it’s own range of incoming cv that will output a single pitch. This range extends halfway to the next chosen note. This is visualized excellently in the built in quantizer on the ER-301. The result of this is that the range of possible incoming pitches varies from note to note depending on the scale/notes you choose to quantize to. If you send a sampled random value to the quantizer the varying ranges means that some notes are more likely to play (the ones without many notes near them) than others.

I first noticed this when trying to send a linear ramp through a quantizer and into the Rings pitch input. With no trig input the rings will trigger whenever the pitch changes. I noticed that the pitch changes didn’t happen at a regular interval like I expected. Another result is that arpeggios will often strike the same note multiple times if there are large gaps between chosen notes.

It would be very nice to have a quantizer that evenly spaces the note regions, I think it would be more predictable for generative patches, especially if you are using basic waves as inputs.

Another related thing I have been thinking about is transposing quantized values and inversions. The idea would be to fold quantized values back around an octave down or up like in a chord inversion. That way instead of getting a simple transposition you would get more varied patterns. I think it is achievable with a sample scanner. I imagine the sample would look like a sawtooth wave with as many cycles as octaves in the range.

This isn’t really a feature request as this isn’t specific to the ER-301, but it has been on my mind and this community usually has interesting ideas when I have quandaries. If you have any techniques for varying generative patches, or any general tips for quantizer use in general it would be very much appreciated.


This is one of the first ER-301 videos I ever made. Even though the scale quantizer largely eliminated the need to do this, I left the video up because I thought it still had a couple interesting things to offer. I think creating your own quantizer like this could potentially solve the linearity thing you’re talking about. It is also interesting in that you can repeat notes and make them more likely to occur than others.

Note that in the video I’m sending some negative voltages into the slice parameter which doesn’t make any sense. What can I say I was a newb. :slight_smile:


This smells like an intriguing thread. Thank you!
Just want to quickly add two things.

  • my general approach to deal with the phaenomenae described above
    Would be to add/mix a second (or more) mod source(s) to the one that
    Is not behaving. E.g. try to add a slower sine to a ramp.
  • I found several modules outside the er301 to be very helpful in this regard which probably can be remodeled to some extent inside the 301. Sometimes just a fine tweak of the amplitude and/or curve of a ramp can yield what you are looking for. Think attenuators and - verters. Ornament and crimes are powerful sources for a wast diversity of randomized or periodical modulation AND quantization.
1 Like