algorithmic composition

I'm just curious if any of you do any algorithmic composition? Can I hear some stuff you've done?

I haven't made any full compositions, but I've been working on some stuff with the Nord G2 demo that is constantly evolving... wouldn't call it a composition really, but I enjoy making a patch, and letting it run in the background. Its kind of soothing.

So yeah. Show me your algorithmic compositions, experiments etc, and tell me what you used and how you made em.

I don't think anyone here is an algorithm, so, no.

ha

I remember the piece you did using the Fibonacci sequence a long time ago.

I wish I knew more about algorithms; what to make them with, good examples of algorithms, etc. I would probably be more inclined to use them if I had a greater knowledge of them.

I don't know much about algorithms either, I'm just kind of playing it by ear.

wow, you remember that piece?

i did a (drum) sequencer (and synthesis) in pd:

this is what it sounds like:

http://forum.watmm.com/index.php?act=attac...ost&id=2217

fun stuff.

yea, sometimes i do noodle patches in g2. I usually hook up a bunch of oscillators to a bunch multiplexers (and then connect the multiplexes to a multiplexer) and control them with a randomizer. If i'm feeling really ambitious, then I'll usually throw in a bunch of MIDI note sends.

im playing around making big patches in energyxt, it's fun to do machinated stuff in there. like you begin with one really simple basic idea and it kind of blooms up around that. nice way of working, but i don't take it very serious. just messing around.

Ive written a number on my G1, I think some are posted here as well..

Here's a patch that takes two compared Julia fractals and quantizes the output:

http://forum.watmm.com/index.php?act=Attac...ost&id=5056

And here's how it sounds:

http://forum.watmm.com/index.php?act=Attac...ost&id=5078

If you really like listening to algorithmic music btw, I can't recommend the works of Laurie Spiegel from the 1970's enough. The stuff she did in LISP on the Bell Labs "GROOVE" system is absolutely amazing (especially Appalachian Grove).

I have some more stuff too if you're interested.

My music is based off of the dynamic algorithm that is my mind.

I also have a killer G1 patch that sounds like Ocean waves... not really an algorithmic composition, but great to have running in the background. Ill post that up when I get home as well.

Ill post this one thing Im doing with the G2 demo when I get home tonight. I can't install wiretap on my computer at work, cause I need a freakin admin password.

I also hacked together an interesting G1 patch that would randomly select between 8 different event and control sequencing rows. It's a goddamned mess of wires and doesn't leave alot of G1 DSP space for synthesis, but it's interesting anyhow.

I can also now just have these things send "triggers" out to the synthesizers.com modular as well, but Id bet it'd work to randomly trigger an SH-101 or something as well, or even a Reaktor patch watching an audio input.

dont mean to be negative but i think aspirations most people have of making purely algorithmic composition are too idealistic. ten fingers nord patches are the direction i think most people interested in algorithmic composition should go in

I personally think much more interesting results are achieved when semi random objects are used to control instruments or even interact with eachother in unpredictable ways. In reaktor or a program like bidule its very easy to do. Having something like a 16-step sequencer's direction controlled with white noise, smoothed noise or even a sine wave LFO (and on top of that the lfo's rate controlled with a random value) can creating very useful controlled random patterns, ones that can slowly evolve over time due to phase shifting but still keep time.

honestly the purely end all algorithmic music composition process is very high level. Unless you are an expert programmer like David Koep i think most of the time the results will sound much less than impressive.

in a similar vein to what i describe above check out the approach robotics expert Rodney Brooks takes when trying to recreate natural insect movements . Here is a paper he wrote about how AI designers are working in the wrong direction http://people.csail.mit.edu/brooks/papers/elephants.pdf. In a movie i saw him in, he spoke about how he uses simple sine waveforms going out of phase with each other to recreate a realistic centipede movement. The movement was a controlled chaotic script that sounded almost like a weird synth patch.

oh, I agree totally. I'm not hardcore into this, and I don't want to make some crazy super algorithm wank shit. (trust me, I'm surrounded by enough of it here, that I know whats shit, and whats just cool).

Nothing that I am doing is high level algorithmic composition. I don't even know if algorithmic is the word really...

the patch I am going to upload later (forgot to do it last night) is basically 3 LFO's that are modulating each others rates. Their outputs are then being quantized to note values that I have chosen, that crossfade with other note quantizers giving semi random melodies. The LFOs are also trigger drum samples, and various synth parameters, so that everything is kind of breathing together. It doesnt evolve into crazy different directions, but you can let it run for awhile, and new things always pop up. What I am actually doing is infact cutting up a 10 min bounce of what the nord is doing, and them I am going through, and cutting out what I like to arrange. Im a big fan of intent. I think algorithmic ideas can be useful and lead you in interesting ways, but I really have no interest in completed pieces that grow completely on their own. Infact, I think that the lack of intent in electronic music these days, is why there is so much shit. People think they can assign some random values and cut up beats with fractal generating max patches, but it sounds like shit cause you didnt say anything with it... you merely brought it into existence. If more people sat d own, cut up the random crap, listened to it, found parts they like, and rearranged them so there is intent behind it (allowing for expression and coherency) then thered be a lot less shit electronic music these days.

imho.

but sometimes I think its just fun and soothing to let a nord patch run in the background. its not a piece, but it is amusing. thats what I was getting at and holy shit this just turned into de ja vu.

Indeed. LFO's are alot more musical than most algorithms you'll be comming up with. I used to have a piece up called "Gently" which was a sootheing portamentoed grouping of LFO's interweaving and comparing to one another. Note/Key quantization is also a must. Without being very precise, microtonality with these sorts of things gets way too chaotic and nobody really wants to listen to it then.

I'm currently formulating ideas in my head for an algorithmic perl script that will use several overlapping iterative algorithms to output a Renoise xml file that is a composition. I could probably generate sequences and songs that sound almost oposite of the algorithmic compositions we've seen in here. By just sticking to simple rules, and heavily controlling some randomness, you could make some really boring safe sounding music that sticks to a key and has a beat, no problem. Once I get time to work on it Ill try and post some results.

Sorry I didn't get a chance to post any of the other patches yesterday, I was busy getting my car out of an inpound lot because I left it on the wrong side of the street during the snowstorm :P

in the patch I made, all the cool variations are made by using different LFO shapes. the rates remain the same, but just changing the shape gives you new rhythms and melodies and effects.

Also, I forget the name of the piece, but one of the lovliest "algorithmic" parts Ive ever heard was on the OHM+ "Early Gururs of Electronic Music" compilation. There was a program running (perhaps on GROOVE), that would listen over a microphone to a piece played by flute. It would respond to it either in key, or modulating to a new key, and then the floutist would respond in kind, and again and again. The interweaving of man and machine in that way was just perfect I thought.

  Kcinsu said:

in the patch I made, all the cool variations are made by using different LFO shapes. the rates remain the same, but just changing the shape gives you new rhythms and melodies and effects.

I wish I could vary LFO shapes on the G1 :(

I kind of could, actually.. using the smoothe module and a square lfo.. hmm.....

Also:

Try having a Saw and a Ramp running against each other (either one rate modulated by the other, or the two modulating oposing pitches, etc). The way the saw and ramp go rushing toward homogeny then rushing away from it makes interestingly musical results.

oh, Im not even talking about morphable waveshapes. I just mean different combinations of waveshapes between the LFOs produces interesting results.

ok, here is a short clip that I just let run free on the G2 demo, I recorded it into logic, and added a hint of EQ and reverb.

kcinsunord.mp3Fetching info...

so, it has its cool moments, but there is no intent. so Ill go and chop up the good parts, and make something kickass with it. or thats the idea anyways.

Wow, that sounds really nice!

Can you post the patch / a screenshot? I'd like to play around with a rework of this in my G1...

thanks :)

blam:

Im not using the FM in that patch, as you can see... I hooked it up, but the mod amount is down. To get different patterns, just choose different LFO shapes, and also change the range of the note quantizers.

Interesting

What are the pitch controls for on the Nord Filters? like pitch tracking a self-oscilatoring filter? I don't gets it!

This is neat though, Ill have to be sure and post some of my G1 experiments this morning. I think you'd appreciate "the Sequenator" (the "randomly chooses 8 control/event sequence patterns that are user programmed)

just the filter cutoff

oh, lol, I thought there was a seperate knob for that, lol