Using Euclidean/Bjorklund Algorithm for rhythm generation entirely in Max
Hi guys,
I've spent the past three days in reading articles and websites about Euclidean and Bjorklund algorithms. I am really trying to understand the algorithm and the calculation that needs to be done so I can recreate it entirely in Max. I would like to avoid using Java scripts, C++, Lisp, etc since I am less than beginner in these, and my goal is not to simply create something that works, but also understand it and learn from it.
Most of the articles however (like this https://ics-web.sns.ornl.gov/timing/Rep-Rate%20Tech%20Note.pdf and that http://cgm.cs.mcgill.ca/~godfried/publications/banff.pdf) are seemingly written for programmers and I'm having tough times trying to understand them.
I've also checked for similar topics on this forum, but the ones I've found are based on Java scripts and C codes.
Does anyone have a good reading, which can help me understand the calculation that needs to be done? Any advice would be of help.
I would really appreciate it.
Thanks in advance!
have you looked my polyrhythmus? it should serve you all the building-blocks you need and the device is free (+ obviously open to look at the code) .
edit: the device is free of javascript and anything alike – it's pure max.
PR is fantastic, btw. If you're looking for an abstraction that wraps around the bjorklund js... well... I happen to have such a thing. it's really easy to use, too.
Let me know.
It has a few features that let you sequence it straight away, too.
my euclidean algorithm max abstraction
http://www.11olsen.de/code/max-objects/11euclidean
oh why did I miss this? You're generating the sequence without js! that's awesome. Can I steal it for my thing? It'd basically be a wrapper for yours. Cheers.
of course, it's shared to be used by anybody.
Guys, thank you very much, I really appreciate it. You're the best!
Is it also possible to share the theory you have used?
I would like to learn from this exercise rather than simply copy-pasting your work.
Thanks again!
i just used this piece of code and tranlated it to max:
for (i = 0; i < len; i++)
{
pattern <= len)
{
SET_BIT(pattern, 0);
cnt -= len;
}
cnt += pulses
}
given and very basic: we are talking about a sequencer. this sequencer has different lanes / tracks (each sequencing e.g. notes, a drumsound... whatever). think of one "lane" as a circle and think of it's steps as equal slices of this circle. the classical (eg. roland) drummachine would therefore have x circles (1 for each of it's individual drumsound). each of the roland's circles has 16 slices. steps. the completion of one circle – of step 1 to 16, is also one cycle.
the difference between polymetric and polyrhythmic algorithms lies within the division of these circles. a polymetric sequencer will optionally have less or more steps within a sequence, anyhow each step's duration – it's meter – will be identical amongst all sequences.
so: decreasing the quantity of steps in a polymetric sequence will make it cycle (complete) faster.
let's say we have a (polymetric) sequence of 8 and another of 3 steps – i'll see if this alignes correctly:
seq1 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 ..
seq2 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 ..
within a polyrhythmical sequence, circles / cycles of any subdivision
can have an equal duration. therefore, you can complete two parallel cycles – one with
3 steps and one with 8 steps – in the same timespan.
in practice and for my knowledge, this is ideally (only?) achieved with a high-resolution clock. a midi clock only serves you 24 ppqn (pulses per quarter note), but to decently divide a clock (because this is what we want to do per sequence) we need a higher subdivision of time – here come the "ticks" into play. each quarter note is divided (i think!) into 1920 ticks. now you have a lot of possibilites to device. see attached a quick patch.
*edit: code-bug cleaned
a bar is 1920 ticks, so it's 480ppq, normally. I don't even know why [transport] reports ppq, it's always 480, it seems.
sounds complicated. that is why i just use metro and the 1 ms sheduler as basis.
why do you guys always work in ppq for a realtime application?
because that's what most transports spit out (:
btw, in my polyrythmic algos i just work on the metrical layer (because mathemtically it is the same thing).
for example if one channel is producing 8 beats at 120 bpm (= 67.5000 ms), i only have to calculate how fast channel 2 has to be to produce 9 beats during the total period of 500 ms.
sounds like an interesting approach, although i'm not sure if i fully understand it.
are you literally using different clocks? don't you're sequences run out of sync when
changing the divisions? would love to see a small example!
different clocks, yes, of course. there is no reason why a bunch of metros would start to go out of sync.
example, yes, but not before sunday.
ppq is a good thing because dividing down from it breaks down into nice pieces for musical timing.
The main thing isn't "how far apart are the beats", but the sync itself. I may have a beat setup where they're 67.5ms apart, but *where* on the beat is that? So syncing to a central hi-res clock solves a lot of these issues.
Also, it allows for neat offsets and other goodies that otherwise take extra patching, I find.
btw, this is the new edit of aw.bjork, with 11olsen's code wrapped up:
i have milliseconds in the code, and BPM, signature, and fractions/beats on the GUI.
maybe i am missing something fundamental here, but what could be the difference of applying the rule of three to ppq as opposed to ms and bpm?
oh you could definitely break down a high-res, central count of 0-whatever in whichever time subdivision you'd like. I *prefer* to use ticks, as it doesn't *need* to have the tempo factored in by me to make sense, it doesn't care. It's the honey-badger of sync.
...to be clear, I prefer it, since it's not a "world-time" subdivision, but a "musical time" subdivision.
The only problem with ticks is that you have to deal with ppq depending on the application. For example, Max uses 480 while Logic uses 960. A more general unit for musical timing would be "beats" as float numbers. In fact, it's already used from plugsync~ in M4L and midi clips in Live. And IIRC, the musical timing of Supercollider is completely based on beats.
Yeah, I will admit that ppq is really idiosyncratic as a remnant from hardware. My application is only ever Max (480), with the now thankfully rare exception of Live, which is... 480. :)
Beats as float would work as well. My code would run on it, too - just [* 480.] in front, and you should be good to go.
I don't know if you're still interested in but I've just published a Javascript Universal Module implementing the Bjorklund algorithm in order to generate euclidean rhythms.
I needed a well tested implementation so I decided to implement it myself.
It is licensed under the MIT license.
Hope it helps,
Adriano
Hi Adriano,
I was actually looking for the mathematical algorithm which I could implement in Max by myself.
Trying to mod Bennys patch, to trigger a click object.
My problem is that it triggers it more than once, sounds like a drumfill/step.
What could be the problem?
hello encoder,
dig into my plugin "POLYRHYTHMUS". you'll most probably find all you need in the patch :)
:D
I would like to "use" it only as an influence, have my own ideas for a diff. device, but this thing just puzzling me :(
Hm.
The output is int. Why can't a simple counter understand it as integer?
hello again, encoder..
have you solved your issue?
"The output is int. Why can’t a simple counter understand it as integer?"
– i don't understand what you mean by that.. can you explain or post a code example?
Hi Benny,
Sort of, It works like a charm.
Built a 4 channel polyrhythmic trigger generator for the modular paltform, makes awesome sequences.
Thanks for the inspiration and for the example!!!!
hello encoder,
glad to hear :) how ever – i'd be interested in knowing how you solved the problem with the float -> int conversion. i remember having had a struggle with this and not being completely content with my solution. would you mind to share (at least that section of your code)?
i'd be gracious!
benni
Actually, i did not solved it, Jus removed the /100 object, and feed the counter with my own clock generator what Im usually using (snippet).
It seems Im limited to 16 steps...
Hello
I'm trying to implement "POLYRHYTHMUS" (Benny's patch) direct in Max Msp. I don't use Ableton. but always crash if i try to open it direct in Max Msp. Do you have a "POLYRHYTHMUS" version only for Max Msp and not maxforlive. Would be nice.
Thanks, and very interesting conversation here
Tried, and I can open the version downloaded from maxforlive.com without problems...
@11OLSEN Wow! This no-js version (bjorklund_11olsen) is really mind-blowing. I have always builded the Bjorklund algorithm by coding. I did not think that this algorithm could be constructed so efficiently in any data-flow environment...
Yes [uzi] is our friend to recreate these for- and while loops of text based code.
Also just want to say thanks @11OLSEN - really useful patch, thanks for sharing it.
Yes thank you @11olsen
This is very useful to me.