Thread Rating:
  • 1 Vote(s) - 4 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Stochastic Generative Mode
#1
Perhaps something like this feature already exists in a firmware I missed, but I don't remember seeing something like this:

The NDLR could benefit from being able to generate melodies in a way similar to the Eurorack module "Stochastic Inspiration Generator".
Here, you have twelve sliders, which have multiple functions. Primarily, they set the probability of a note occurring in the generated melody.
With further functions, you can set the predominantly played relative octave per note (from -2 to +2).

You additionally have potentiometers for the probability per note-duration, the global probability per octave, a "linearity" parameter that controls how likely the algorithm is to select adjacent notes (still weighted by the individual note-probability), a parameter each for the likelyhood of more ascending or descending steps, one for ratcheting probability, one for rest-probability (where first a note-duration is generated from the given probability-distrubtion, then a random binary note-or-rest value is generated from the set rest-probability), as well as "retroactive" loop controls, which capture the last 18 events and loops a selectable range of them.

Not all of these wonderful features would need to go into NDLR, though the main ones would be very lovely. I can imagine two modes where you either set the note-probabilities for absolute notes (where you'd have exactly twelve sliders/bars) or 5-12 for relative scale-degrees of pentatonic scales up to the chromatic scale.

The UI of the NDLR should be pretty well suited to implement such features - all the parts are there. Speaking as a software engineer, some of the algorithmic details here of course aren't trivial, but they're also not crazy complicated.

Would be an amazing added value IMO. I'm getting a lot out of the Stochastic Inspiration Generator in Eurorack - it's only disadvantage is the lack of easy transposition and modulation (you have to change all the sliders). NDLR could improve upon that with in-scale probabilities and its transposition/modulation capabilities.

I think this would make it one of the most formidable hardware generative sequencers on the market, if not *THE* most capable one. SIG would take that crown now because of its unparalleled fine-grained control over so many probabilistic parameters - were it not for the lack of in-scale transposition and simple modulation.

Thank you for this great little box I've enjoyed so much over the years!
Reply
#2
I like the idea. It sounds like something that I would like to use too. It isn't likely that we will be adding much more in firmware at this point as the device memory is pretty full, but it does sound like a good time.

In terms of what is already available in the NDLR fw to get generative patches going, in the LFO menus there is an option to set the LFO shape to a pattern number. I have found these to be really rewarding to route to elements of the device like the selected chord degree or chord shape, clock divisions etcetera. You just want to be sure to experiment with it to make sure the LFO time is longer and clock synced so that it tends to stay rhythmic (or not if that's your thing).

I wrote a little more detailed post about this technique somewhere, I'll see if I can figure out where and link it for you here.
Anyway, thanks for the suggestion!
Reply
#3
https://conductivelabs.com/forum/showthr...ve#pid2190
Reply
#4
Although I really like the SIG, when I read the OP I thought this was getting away from the initial intent of the NDLR. I thought of an alternative use of the NDLR to kinda sorta approximate some of the features of the SIG. But, I didn't want to come off in a negative fashion, so I let it go. Then Jessie came in with his enthusiasm and suggested the LFO pattern possibility for generative stuff, which was encouraging.

This morning I decided to interrupt my working on my current song to try out my thoughts that initially came to me while reading the comments above about the SIG. My gawd, this thing is beautiful and exactly what I was looking for since my current song is heavily random and probability based. Here's what I did:

To simulate the note probability sliders of the SIG, I used the Pattern Editor to set probabilities of notes by repeating entries of a particular note. Then, set the Motif's Variation to Random. VOILA! Probability of notes desired is the result. For example, the pattern could be 0-0-0-0-1-1-2-2. This will yield a probability of note 0 playing 50% of the time, note 1 playing 25% of the time, and note 2 playing 25% of the time. Of course, this has nowhere near the flexibility of the SIG, but it kinda sorta gets there part way. The Pattern Editor is limited to a grand total of 16 notes, limiting the granularity of probability - especially with greater numbers of total notes to select from in the desired pattern. But for my needs, where I only wanted five notes to be used, I got just the amount of weighted randomness I was searching for. Plus you get the great transposition due to chord degree changes that makes the NDLR so intoxicating. Thank you #Majickthise.

By the way, the pattern I ended up with was: 0-0-0-0-0-1-1-2-2-2-3-3-3-4-4-4. Which results in 31.25% note 0, 12.5% note 1, 18.75% notes 2, 3, and 4 each.
Reply


Forum Jump:


Users browsing this thread:
1 Guest(s)