#42

Now we’re talking. If I count correctly, the Erica Synths has 29 outputs counting the CV/Gate pair in addition to the triggers and accents. I knew it could be done.

#43

And how do you go about applying such tone row manipulations on the modular? I’m very curious.

That’s a good suggestion, thank you for bringing that book up. Any success in your application of his techniques on the modular?

#44

I wasn’t aware of the gamelan angle, but this is what tends to happen in self-generating Krell patches with smaller configurations - if you apply the core random CV value to v/oct as well as the decay, note length is naturally tied to pitch. With Maths, at least, higher CVs will lead to shorter notes by default. It doesn’t factor in lower notes playing less often, but this is easy enough to configure with the right module.

I don’t know a great deal about gamelan music, though the uni where I work used to have one installed near my office, so we’d quite often find ourselves working to the sound of student gamelan explorations. I’ll have to listen to a few performances and see if the Krell is a good starting point for an emulation of technique…

2 Likes
#45

I’d love to hear more about this!

#46

It also mimics acoustic phenomena - higher frequency strings, eg, will decay and ring out earlier. It’s usually not a drastic difference, but it’s one well-established in our perceptions, and we appreciate hearing it in synth patches (in my experience, at least).

Edit: of course, you were replying to @eblomquist, who already said this.

2 Likes
#47

The strength of modular is in creating generative rhythms. For this we don’t need that many different sequences. We can derive many related rhythms from a single sequence.

If you just want to program rhythms, it’s probably more cost-effective to pipe it into the modular via midi.

1 Like
#48

The basic serial techniques are:
forward—the basic sequence
retrograde (reverse)—most sequencers have this feature. It’s CVable on the A155/154
inverse—invert the melody CV coming out of the sequencer
inverse retrograde—invert the reversed melody CV
retrograde inverse—you’d need to program it into a second sequencer
transposition—mix CV in with your sequence
use subsets of the row—the A-155/154 makes this easy, by CVing the first and last. Another fruitful tactic is to S&H the original sequence to get new variations.
rhythmic variations—use a gate sequence as your melody sequencer’s clock; or CV gate length on your melody sequencer; use different clocks/rhythms to trigger your S&H, etc.

By combining the above, you can get endless melodic variation out of a single sequence.

Just for starters, try this—use a gate sequencer—Grids, Zularic, whatever—as your sequencer’s clock. If you don’t have a gate sequencer, try your drum machine’s audio output as your sequencer clock. (You may need to amplify the audio.) Then mix your sequencer output with an lfo before quantizing.

—————

Johnson’s techniques are mostly trivial to implement in a computer programming environment like Max or PD, but patch programming them in a modular is far more challenging. I’ve figured out how to do some of them in modular. It’s a fun mental exercise for me.

You may enjoy the articles on Johnson’s website. There’s also a cool blog on “algorithmic composition”.

14 Likes
#49

So much good stuff in this thread.

The thing that keeps coming to my mind - and that, frankly, I am feeling an urge to write a Norns script for when I can get me a Norns - is change ringing as a model for ‘tone row manipulation’ - that is, the traditional English practice of bellringing, in which a series of fixed pitches are cycled through rhythmically, each bell controlled by one person, and the order of rings changing according to an algorithm each cycle until you end up back where you begun (important so all the bells are hanging ‘up’). As one programmer friend realise when he was trying it, it’s basically sort algorithms. Here’s a diagram of eight pitches being cycled through from wikipedia:

Call_changes_on_eight_bells_WQT.png1138×1948

Patching this on modular: might be doable depending on sequencer. Feels very straightforward for Teletype/Norns; I think what’d be interesting is finding UI to play this - after all, changes often change sort-algorithm as they go.

10 Likes
#50

I know nothing of change ringing, but in the classical tradition this approach is called hocketing—where a different instrument plays each note in a melody. It’s easily implemented in modular with a sequential switch. Melody cv in; outs to different voices.

2 Likes
#51

It’s not quite hocketing - as well as “each note played by individuals”, the point really is a repetitive pattern where the order of notes shifts on each repetition. The latter is more important than the hocketing.

This is nice:

on explaining the physical limits on change-ringing. Also, footage of people doing it.

4 Likes
#52

Thanks for the correction. After reading the WP article I’m still a bit confused, but the changes remind me of a technique in Indian classical music called “Merukhand”, where the musician performs all possible permutations of a pitch set.

1 Like
#53

oh! yeah, that’s really, really similar - that cycling through all permutations of a small combination. I didn’t know about that!

#54

I think the difference might be that change ringing is necessarily mechanistic due to the setup—multiple musicians, considerations of whose bell is up versus down, etc. Merukhand is usually implemented by a single musician who uses the permutations as fodder for improvisation.

The improvisational approach could easily be implemented in modular with Pressure Points. Each touchpad is one note/voice. The performer works his way through the permutations freely. Idea for a piece…!

#55

Thank you all for sharing your sequencer ideas and preferences! This makes for very interesting reading. Given the discussion around producing variety from a small amount of material and some system of rules I thought some of you might be interested in a Eurorack sequencer prototype I’ve been working on.

It’s a grid-based sequencer called STATION which borrows the ansible paradigm i.e. it will eventually take the form of a grid-controller e.g. a Push or Launchpad (non a monome grid unfortunately), that is hosted by a module.

The reason I mention it here is that it has a permutation algorithm that makes a mathematical model of your patterns as you programme them and then uses that model to make permutations on the fly on playback. The amount of permutation can be adjusted from 0% to 100% at any time.

Below is a link to a recording that gives you some idea of how that might sound. First you hear an 8-bar arrangement played with 0% permutation. Then after a short gap there are a couple of minutes of playback with permutation set to 50%.

You can hear ‘the one’ slips around, certain parts of the arrangement repeat but nothing is ever the same twice.

https://soundcloud.com/dennismcnulty/station-improvisation-demo

5 Likes
#56

I like the Dilla-esque “don’t cry” treatment. Worth looking at Pucktronix Oracle, not the same as what you’re doing but in the same family.

1 Like
#57

Thanks. Glad you like the tune :slight_smile:

A friend mentioned The Oracle to me a while back. There is definitely a similarity in the sense that it looks like that module also makes use of Markov Chains. But there are a few differences.

First difference. There’s no learn phase. STATION can immediately generate permutations because it is integrated into a sequencer. Making this work with an intuitive interface is a big focus for me. The grid-controller layout is well understood. I want to make a device that can slip smoothly from deterministic playback to permutational playback and back again.

Second difference. The way I’m using Markov Chains is, as far as I can tell, unique. I haven’t come across any other examples of anyone using the same approach. I have searched!

Third difference. It applies the permutation to 8 channels simultaneously. I’m only using 5 in that track. Each channel can be a trigger, gate or cv output.

Fourth difference. It also just works as an 8-channel sequencer without permutation.

7 Likes
#58

If you have access to a Maths, Tides or equivalent, and some kind of CV-friendly synth, you’ve probably got all you need (and VCV Rack / Reaktor modules will also have you covered). I’d recommend looking up Krell patch videos on Youtube, but in terms of my post the core of the patch is pairing a random(ish) CV with the output of a function generator. You send the CV to your oscillator to set the pitch, and also to your function generator to set the rise and/or fall of the function. You apply the function CV to your VCA, and that’s the basic patch. Every note will have a different pitch and a different envelope length, both set by the random CV value. The end of fall gate on the function generator triggers a new random CV value and also retriggers the function, so the end of each note creates the next note.

Because Maths is set up to create shorter the rise / fall times as the ‘both’ CV input increases, you end up with shorter envelopes for higher pitches, so your patch tends towards long bass notes punctuated by trills of higher notes.

Obviously without applying some bias to your random CV, long sequences of bass notes are a possibility, so some tweaking is usually needed there to make lower values a little less likely. And there’s plenty of scope for complexificating the patch with filter modulation, quantised CVs, giving the pitch and function separate CV sources, etc. etc. But once you’ve got that core setup of a function generator, VCA, random source and oscillator, you’re up and running and ready to head in whatever direction takes your fancy.

Hopefully that’s of some use - apologies if it’s grandmother / egg information.

9 Likes
#59

This is very interesting. Thanks for sharing! I’m very curious to see the UI for this one.

#60

Thanks, this is great!

I have a Maths and a Wogglebug as well as an Optomix /ModDeMix so I should be able to rock some Krell on the basis of this information! Psyched!!!

1 Like
#61

[quote=“Nick, post:58, topic:25375”]
Obviously without applying some bias to your random CV, long sequences of bass notes are a possibility, so some tweaking is usually needed there to make lower values a little less likely.

Is bias possible using a Maths channel 1 or 4?