Random Rhythm Generator - Probabilities

w minor · Joined: Sep 28, 2006 Posts: 12 Location: Reading, UK

Hi,

I'm in the process of programming an algorithmic composer that will generate rhythms as midi signals.

A random number generator feeds into each patch and some sort of probability determins weather or not that number triggers a note or not.

I would like to give these probabilities some sort of mathematical significance rather than just random values.

Something to do with the golden ration perhaps or the fibbonachi sequence? i haven't studied maths for years so my memory is a bit rusty and i'm having trouble figuring out which way to go with this particular aspect of my project...

i know this isn't really the place to ask but i wasn't sure where else to look... so anyone with any ideas or who could direct me to some starting points it would be very much appreciated.

Thanks

w minor

Kassen · Posted: Sat Sep 30, 2006 11:17 am Post subject:

You might want to have a look at breakage for inspiration:
http://www.blackholeprojector.com/

dewdrop_world · Posted: Sat Sep 30, 2006 11:36 am Post subject:

Wow, breakage looks really cool.

BTW, 2 of the people listed in the "thanks" section are avid SuperCollider users (Nick Collins and Thor Magnusson), both of whom I had the pleasure to meet at the sc symposium in Birmingham this year. Nice chaps.

I'm also doing algorithmic rhythm programming in sc, will share some thoughts later. Right now I've a small project to try to finish today, so not much time for chitchat...

hjh

Kassen · Posted: Sat Sep 30, 2006 11:53 am Post subject:

Cool!

I'm not so much into litteral algorithmic rithem creation but lately I've been quite deeply into using algorthems to modify hand-programed rithems (X0X style) based on live input (think breakdowns/ fill-in, etc). this is in ChucK.

Anyway, no time for me either tonight (I have to spin some records) so let's share notes later!

kruhft

w minor · Joined: Sep 28, 2006 Posts: 12 Location: Reading, UK

funny you should say all that, but that's exactly what i've been doing anyway. I decided to get my head round just building the thing before i worried about stuff like that. No time to fully explain now but will be backs soon to explain and post the patch

laters

kruhft · Posted: Sat Oct 21, 2006 7:11 pm Post subject:

"Make everything as simple as possible, but no simpler" -- some guy that said that Very Happy

dobart · Joined: Feb 20, 2007 Posts: 10 Location: SEATTLE

There are some interesting 'candle flicker' circuits that generate nonrepeating random signals; IE a 2mhz oscillator being sampled by a ~100hz oscillator through a flip-flop. I breadboard these to control vactrols to provide 'random' variation, and the samplng oscillator can be clocked to give discrete pulses or higher frequency noise. So far I've used 3 stages of this circuit in a 3-color LED lamp, which drives my LDR controlled instruments in cool ways. Another source of randomness can be feedback, where the output of the instrument feeds back into it in complex ways. I played a bit with that on this LDR controlled 556 (APC) +741 wsg VCF thingy, breaking the CD4040 and LFO driven LED feedback up through LDR gated oscillators:

http://www.youtube.com/watch?v=VGsdvYE9W4E

here's the 'flicker' circuit

http://members.shaw.ca/novotill/FireLightFlicker/

williamsharkey · Joined: Jul 31, 2005 Posts: 61 Location: usa

Markov chains are a good way to organize random jumps.

Even though they are called "chains", I think you should forget about the word "chain" for now. Try to visualize frogs jumping from lillypad to lillypad, that's what markov chains are all about.

Lets say there are two states: Snare, and Kick. Now, we program the machine with example probabilities:

After a kick, there is a 100% chance of a snare.
After a snare there is a 50% chance of a snare, and a 50% chance of a kick.

Here is what the Probability matrix would look like:

williamsharkey · Joined: Jul 31, 2005 Posts: 61 Location: usa

jksuperstar · Posted: Tue Dec 18, 2007 11:51 am Post subject:

The markov chains could also be note lengths, or multiple things triggered by an event...after a whole note bass drum (or bass on 1 beat), you can have probabilities that intertwine the probabilities of other instruments. 50% chance of snare, 75% chance of hi-hat 16th notes, etc. Or have 2 markov chains triggering the same sound, so you get the complexity of 2 patterns mingling (and since the sound triggers, one pattern effects the other!).

frijitz · Posted: Wed Dec 19, 2007 7:42 am Post subject:

I'm exploring the use of chaos generators. Chaos in the correct mathematical sense, not the common sense of "random". Chaotic signals are deterministic but irregular. The amount of irreguarity depends on the mathematical system and its parameters. I have some simple demos of results from some hardware I have made over the past few years on my website:

http://home.comcast.net/~ijfritz/ch_over.htm
http://home.comcast.net/~ijfritz/ch_close.htm
http://home.comcast.net/~ijfritz/ch_circ.htm

Very Happy

Ian

Inventor

plagal · Joined: Dec 28, 2007 Posts: 18 Location: South Carolina

Just to add my 2 cents -

Maybe have a chain of a few arpeggiators that do switching in various ways such as a tempo change every-other time an arp plays. Same for delayed notes, overlaping, rests, muting, and number of notes each arp plays in turn, and so on...

There's nothing random about this, but its cool to get accurate results from playing with the controls.

BTW this has already been done, but perhaps someone can do it better -
but then, since it was me who already did it, I doubt it Very Happy

Undergrind · Posted: Mon Jan 28, 2008 12:04 pm Post subject:

Haha

i just posted somthing about that here: http://electro-music.com/forum/topic-23494.html

without knowing this thread..

maybe melt them together??