Generative Systems

category: general [glöplog]
betcha i can make generative music that sounds cool ;)
added on the 2007-08-23 20:19:56 by Gargaj Gargaj
I'm pretty sure you can, but not without specifying such strict rules for how the individual parts of the music are to be generated that it's borderlining on sequencing/tracking anyway. :)
added on the 2007-08-23 20:26:08 by gloom gloom
it's not like you dont have the standard "intro, chorus riff, verse, chorus, verse, chorus, variation, solo, chorus, outro, coda" structure in music anyway. ("intro, buildup, hook, repeat, breakdown, hook, outro" in electronic music)
added on the 2007-08-23 20:34:43 by Gargaj Gargaj
Yeah, but you would probably need to define some ground-rules for how those sections would be generated, and also set up a different set of rules for how the individual parts of those sections would need to be generated. For example; melodic bass-hits shouldn't collide with BD-hits, the correlation between hihats and bassdrums, how the melody should run alongside the bass... and stuff like that.
added on the 2007-08-23 20:40:02 by gloom gloom
once you figure out what genre you want to rape, you're pretty much set... for one example, hands-up-in-the-air trance doesnt really require much brain to write, even less to generate.
added on the 2007-08-23 20:45:38 by Gargaj Gargaj
Oh Sir, I take offence at that statement. :) I have written my fair share of "bangin' trance" back in the day (still do actually, but I'm _never_ telling you my alias) - hey, money is money.

What is REALLY braindead is hard house - that shit writes itself in its sleep. In a bag.

Under water.
added on the 2007-08-23 20:48:13 by gloom gloom
added on the 2007-08-23 20:48:56 by Gargaj Gargaj
Oh yeah, my point; BAD trance is easy as hell to write. GOOD trance is just as hard as other types of music.
added on the 2007-08-23 20:49:13 by gloom gloom
*hides the JP-8000*
added on the 2007-08-23 20:49:44 by gloom gloom
hmmm, houdini is still an iso.. should take it for a spin :D
Gloom isn't it exactly the same be it graphics or music? You have to define rules that work on a wider aesthetical level in both cases.
added on the 2007-08-23 22:24:54 by _-_-__ _-_-__
at this point I'm so afraid overhyping is slowly going to kill spore :/
added on the 2007-08-23 22:30:20 by _-_-__ _-_-__
Brian Eno?! Now I must play the game.
added on the 2007-08-23 22:34:22 by xernobyl xernobyl
knos: I don't feel that way. Music can't be just "blip-blop" (or atleast; _I_ think it can't :), but graphics/effects can be blinking, shaking and random seeds of ribbons (hehe) flying about.
added on the 2007-08-23 22:39:35 by gloom gloom
gloom: But once it's random, it usually looks crap. Just like with random music. To make some interesting generative music would require a lot of control, yes, but so does interesting generative graphics.
added on the 2007-08-23 22:53:39 by kusma kusma
gloom, maybe you think that way because we tend to be more prejudiced when it comes to music than graphics? I mean I still encounter people in 2007 that claim that hip-hop is not music or something silly of the sort.
added on the 2007-08-23 23:49:08 by _-_-__ _-_-__
in any case I think it's rather paradoxical actually to claim music cant be made generatively, when in some form much of what passes for the tradition of western music has an element of generativeness.

Classical scores in a way are just a sheet of procedural steps to follow, which, however, to bring us the real music must be actually interpreted by a human player.

The composer could not have imagined the interpretations we have been given to their pieces..

It's so natural for musician to play with these ideas of rules/games/sharing of control. I mean look at this.. .
added on the 2007-08-23 23:55:03 by _-_-__ _-_-__
I tend to think that generative involves using some man-made set of data (music, graphics, whatever), to extract some set of rules. Then you let the rules run on their own, and some stuff gets generated. It may look fairly nice, and is interesting because it may create unexpected results (although it will always be wholly unoriginal).

markov chains are of some use, apparently. eigenvectors too.

personally, I think that the algorithms behind generative stuff are much more interesting than the end result itself, which will never get beyond a mish-mash of the input data.
added on the 2007-08-24 00:52:04 by Reboot Reboot
However, I intended the thread to talk, not about generativeness as something interesting, rather about generative techniques as something enabling us to realize what otherwise would be too painstakenly complex/deep/wide/large to realize.

systems, letting you create complex visualizations which aren’t possible to create by the normal approach (edit objects manually).

And good examples of that.

As smash pointed out with his comment about media-error, those techniques are very well at hand when creating models / textures off-line.

(By the way, gloom, I guess musical wise there is not much to do here,
unless we for example create the imaginary soundtrack of a going through a city, or better yet something unrecordable like trying to create the sound of a colony of ants ;)

You would think that realtime creation might be motivated by things such as size restrictions. However the trend with intros nowadays is to look like they were made by hand. Which is great, I'm not disputing that.

added on the 2007-08-24 08:11:17 by _-_-__ _-_-__
Knos: We could try to create microblocks of musical elements, assign values to them (intensity, preference, mixing-suggestions) and then write a piece of software that glues it together procedurally to a whole song. Yes, the individual elements would be "man-made", but the song would be unique every time. Stuff like that has been tried before I think.

When it comes to such things in music, people tend to avoid rythmic pieces and just go for long pads and strings blending into eachother. Sure, it's pleasant enough, but not really what I would call music in the traditional sense.

If any coder-heads would like to try something with generative music, I'm game btw. :)
added on the 2007-08-24 08:37:37 by gloom gloom
genetic evolution ftw
added on the 2007-08-24 08:57:59 by mind mind
One could also create a musical mapping of notes; a list of notes that you, the author, find "compatible" for example in a chord or an arpeggio. In addition, a set of progression-rules could be made.
added on the 2007-08-24 09:16:29 by gloom gloom
Gloom, before thinking how, shouldnt we ask first why and what? I've got an interest in generative techniques when i seek certain characteristics: better productivity, more consistency of the result, or because i could not follow the precise steps of an algorithm, or even would like to experience what the ultimate results of a system would be. (modal music for example) It is also intriguing to hear&see how the system leaks out of many variations.
added on the 2007-08-24 09:55:27 by _-_-__ _-_-__
"Why?" is simple to answer - because we are sceners and enjoy these things. :) Also, it could be used as a tool to help musicians struggling in melody hell. Stuck at the bridge? Input your notes and the software suggests some transitions for you. Need a new riff? Input your basenotes, tempo and desired intensity and the software comes up with something (MIDI output perhaps, so it can be tweaked and reused with a whole bucketload of different softsynths).
added on the 2007-08-24 10:03:42 by gloom gloom