What I would be looking for in a generative sequencer is the ability to not only make variations on patterns/sequences, but something that really evolves the input. So that I give the parameters it should start out with, but the sequencer takes it some place I wouldn’t have landed myself. Preferably with multiple tracks.
I have the Oxi One and like it very much, but I struggle to take it beyond variations of the same thing when using lfo’s and the mod options.
Maybe the Conductive Labs NDLR. I only owned one briefly and can’t remember much about it now but for evolving textures and ambient it might be something that interests you.
Owned that one a couple of years ago. I found it very good at chord progressions when playing yourself, but it was a bit cryptic for me when I wanted to automate things at a proper level.
I think this one comes the closest, from the sequencers I know. But it’s monophonic (afaik).
Also I am not able so far to automate things to get chord progressions. Maybe that’s possible, but I haven’t figured it out yet. Also: I do not understand why euclidean sequences cannot evolve. Why, for example, can’t you automate pulses, length and rotate on most sequencers?
I’m guessing you’re looking for a desktop sequencer and not modular but given the topic I have to shout out the Euro module Clank Chaos. It’s my favorite for this kind of thing.
You get 6 channels with gate and CV out. By default, each channel is just spitting out random values with every gate. But you have a set of controls to sculpt the randomness: can quantize it to a scale, set probability, constrain the range, multiply or divide the clock, transpose, etc. until you like what you hear.
At that point, you can loop the outgoing voltage/gate pattern (at a loop length you choose) and start applying all the same parameters to your loop. Probability, transpose, etc. all now work on your loop.
So you can fall into this flow of sculpting, looping, sculpting, looping etc. and can incorporate external signals as well (could trigger the loop externally or transpose with CV). It’s a wonderful way to explore melodic ideas.
There’s also a “chaos” function that applies randomness to all parameters, except you can dial in exactly how much randomness you want to allow for each of your controls. This means you can easily roll the dice on a new melody without the whole thing going totally off the rails. Brilliant design.
this^ - not to be underestimated! per step parameter lock arp speeds and notes, LFOs to notes, arp speeds, transposition, velocities, or CCs, all in key…chain some patterns via arrangement view if you want to further define an arrangement journey
Was going to add, if you set up p-locked lfos on a midi track and then just play whatever you feel like into the OT, when the sequencer gets to that step and grabs your notes and does things to them, all quantized to the scale, its pretty magical. then you can have the thru track of what you are sequencing on the other side doing even more evolving madness
Any chance at approaching this with a Tonverk? (or an A4 MKI?)
That really sounds very cool. I have a modular rig, but I use Oxi One to sequence that and have stuff like Make Noise Multimod and Wogglebug for randomness. But your suggestion sounds like something I could try out some day.
How are you using the LFOs? What about the modulation lanes?
There’s a lot of ways to accomplish this without moving from Oxi one which is one of the more deep generative hardware sequencers you can buy. Using the sample/hold LFO as opposed to normal ones with slower rates than you are now, running multiple sequencers at the same midi channel, and now things like the new domino mode where you can sequence gates separately from notes.
A lot of the more evolving and self playing patches I’ve seen use feedback patching extensively, there’s definitely an art to it, but I’ve also had a lot of success manipulating the oxi’s sequences in real time by adjusting octave range and accumulators, ratcheting/step rate/step direction, and using sample and hold LFOs.
One trick I like is to use low probabilities on mono sequences that have very long triggers, so that if they aren’t interrupted by another trigger the note holds but if they are it ends.
Tonverk has parameter locks on the arp and probability/conditions for those triggers. It also has individual probability and condition per note in step edit, which makes it a lot more powerful for generative sequencing than previous elektrons.
I really will say though, if you have an Oxi one already you have one of the best tools that exists for this stuff and you should focus on learning lots of self generating and evolving patch techniques that are used in modular and figuring out how to implement them in Oxi. A lot of them can be done and some can only really be done with crafty cv/gate patching in virtual or physical modular because they’re not using sequencers at all, they’re using extensive chains of shift registers and clock dividers and alternative patch out points that aren’t just gate and note cv.