|
math+music (algorithmic/generative?)
StoreTags: generative, algorithmic, stochastic
Viewed 3871 times. 18 people liked this blog. You can rate it below if you haven't already.
People who enjoyed reading this: sprouts, astroid, daswesen, bla, sAMsKi, dustmite, lysdexic, djugel, Roshi, timmo, deltasleep, mlbot, nagrom, eyesnine, bew, yghartsyrt, evilai, ignatius
-->
OK, well I thought it would be good if someone initiated some discussion about generative/algorithmic/stochastic music, specifically, since I think it's interesting, but problematic, and there seems to be a lot of misunderstandings about it, and also lack of clarity. I'll admit here that I'm not 100% clear on this subject, either. And also, maybe I'm too late to the party and everyone is already sick of talking about this. Who knows.
First off, I think that people should know that the grand daddy of all of this nonsense is NOT Autechre. Autechre are some nerdy dudes who liked hip hop music but, since they were so nerdy, maybe geek-core polymetric noise fuck-ups instead. Autechre is, I suppose, one of the clearest examples of the intersection of the pop music world with the academic world. But, still, listening to confeld, you still hear those basic drum n' bass, hip hop patterns all the time.
I think I can be safe saying the grand daddy of all of this is the man XENAKIS. Now, personally, I think Xenakis can eat my dick, and I don't really like his music. Nevertheless, we have to give him credit for being the godfather of all this nonsense. If you hate algorithmic music, find a time machine and go back and shoot him in the face. Here is a link to a PDF file of his paper--
Formalized Music: Thought and Mathematics in Composition
link
an interesting read, but pretty dry.
Anyway, this shit has been around for a long time, now. It's nothing novel, except for maybe in the pop music world. If people would stop treating it as a novelty and maybe more like--a tool for making more interesting music, then I think it would be more useful.
From what I understand, algorithmic music, generally, is about the intersection of math and music. I think that to really make interesting generative music, you have to have a deep understanding of both. I think that is one of the problems--so much knowledge is required to be able to do this well.
I was accepted to a semi-exclusive type multimedia art major at the University of Washington that is into this sort of thing, and upon arrival, I found that people were trying to make the future of music and art by poorly hacking together different mediums, disciplines, and fields of study that that didn't really understand. We were trying to make art that was about science, math, or computer science, but nobody knew ANYTHING about art, math, science, or computer science! They were just trying to be some sort of medium between everything, without actually knowing anything. I'm forgetting about that major and thinking about a comp. sci degree--because I'll actually LEARN something. I would do math, but I'm not smart enough, and my wife has a math degree, anyways ;).
Resources about this type of thing? I bought, and have been working through this book, which I think is very good:
The Computational Beauty of Nature
link
It discusses different types of systems, including cellular automata, chaos theory, fractals, and others.
Thing is, the type of mathematical systems that produce interesting behavior are not simple. Chaos theory, for instance, requires a strong background in both math AND physics. Elisa (my wife) didn't have much to say about that. She had a good explanation for me, though, when I asked her about Markov chains, since astroid is always going on about those goddamn Markov chains ;). Even fractals, which are fairly simple, require a good handle of linear algebra, which I don't understand very well right now, either.
Does anyone else out there have some input on this? I'm about 100% certain there are some dudes out on this site who know all about this subject but just haven't jumped in.
Personally, I feel that the problem with generative music is that it leaves so many decisions up to a computer and removes the human element from the art, or the composition. I always thought the human element was the most important part of any art or music. So, in the end, if I end up using these ideas, I have to integrate them into that belief I have.
| |
Comments
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
astroid
i don't give a shit about the label, that's true.
but, i want to make stuff in real time that sound like my composed, sequenced stuff, and i can't think/play that fast.
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
skinnyjabba
sort of on topic
link im not so sure about the whole cheap-soundcard-midi vibe or the nebulous way the hamsters position affects the system, but yeah, the idea of a cage full of animals being an instrument is crazy. with a little work on the algorithm side of things, and a more obviously way the hamsters position/direction/movement effects everything, taht woudl blow minds...
imagine a live setup with like 4 cages of 4- 8 hamsters, each cage set up as a step sequencer with a different # of hamsters/different length, a guinea pig running on an exercise wheel controlling the tempo. and a cat on stage having his way with them
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
jdg
i  the midi hampsters
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
mlbot
i'm sure you meant booty hamsters.
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
astroid
richard gere's aleatoric ass hampster orchestra?
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
mlbot
i cant tell the difference between them and you, astroid.
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
bla
ill try to explain the system i used to program beats on my sr16
it can have 12 sounds in a drum set so for each set i chose 4 low (kicks,toms,congas etc) 4 mid (snares claps,cowbells etc) and 4 high (hats,shaker,triangle, etc) i assigned the outputs so each of the 4 outputs gets 1 low 1 mid and 1 high
i started with a 2 beat pattern and stuck to 16th notes- no weird timings- so theres 8 steps in a pattern
there are only 8 velocities you can enter so i put in 8 hits for each sound- 1 at each velocity - this means theres no gaps but a velocity 1 works like a gap
first i put in the velocity 8's for each sound- i put the 4 high sounds in the gaps where there werent any low sounds and i put the mid sounds so theres 2 on the lows and 2 on the highs
positioning the hits works like this: there are 4 steps to a beat (ill call them A B C and D rather than 1234 so as not to confuse with the velocity numbers) a is on the beat , b is after the beat, c is offbeat and d is before the next beat
after putting in the vel 8's for each sound i filled in the rest of the velocities
if the 8 was on an A or a C then the 7 goes on a B or a D , the 6 goes on an A or C (whichever the 8 want on) and the 5 goes on a B or D (whichever the 7 wasnt on) then the 4 is the same as the 5 but on the other one in the pattern (2 beats = 8 steps so there 2 A's 2 B's 2 C's and 2 D's) the 3 is the same as the 6, the 2 is the same as the 7 and the 1 is the same as the 8
eg: (if the 8 is on the first A)
ABCDABCD
85371462 or
82641735 or
87341265 ....
i avoided stuff that went 8765 or 1234 etc and i tried to make it so that the 7's didnt lie on the 8's of other sounds from the same group (low,mid,high)
as the pattern fills up you get less and less choices
i think i made it so that the highs were exactly the inverse of the lows
eg:
kick
ABCDABCD
85371462
cymbal
ABCDABCD
14628537
errrrrrr i dont think i can explain this properly sorry
anyway that makes very dense 2 beat loops
then i copied it to twice as long (4 beats) and changed the numbers in the second half
i cant remember what it was but it was something like:
8>4
7>3
6>2
5>1
4>8
3>7
2>6
1>5
then i copied it again to make it 8 beats long and changed the numbers in the second half again
possibly:
8>6
7>5
6>8
5>7
4>2
3>1
2>4
1>3
with the patterns as very dense 8 beat loops at 133bpm i played them at my first live set at the first stfu 2 years ago!
but now ive processed them further because 48 8 beat patterns played 4 times each at 133bpm is only about 12 mins
but the sr16 cant really fit any more hits in its memory so to make the patterns longer i had to thin them out
i copied the pattern again to make it 16 beats long then went through it alternating on each step deleting odds (7's,5's,3's and 1's) then evens (8's,6's,4's and 2's) in the first half then the opposite (alternating evens then odds) in the second half
it seemed to drag on a bit so i sped it up to 156bpm
then i copied it again to make it 32 beats long
this time i went through it deleting 6's and 4's and 7's and 1's for 2 beats then 8's and 2's and 5's and 3's for the next 2 beats and alternated it like that (2 beats at a time) until half way then i swapped it to deleting 8,2,5,3 first and 6,4,7,1 second and and alternated it like that (2 beats at a time)
so none of the original pattern data was lost- just spread out
and it worked better when i sped it up to 167bpm
but the drum machine ran out of memory so ive stopped at 40 patterns which is about 1/2 an hour and it sounds fucking ace! better than any beats id tried to write before
hooray for systems art!
-
sorry i expect this post is totally pointless and of no interest to anyone but fuck it- ive done it now
thats how yo make teh bla baetz
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
bla
ive just given away my best secret!
but at least im so bad at explaining no one will understand it
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
mlbot
I thought bla beats were when you used audio-to-midi conversions and made HR16s play themselves???
link
I totally understood that, but this latest manifesto is very confusing.
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
bla
thats the other thing i do
i was talking about when i actually program the beats
youll hear them in a month or so after ive recorded the gig im going to do
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
bla
but theyll be drenched in crazy effects so theyll still not make sense
maybe ill upload some fre lops of the uneffexed drums
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
ricemutt
haha, whoa bla, whoa astroid!
Yeah, I think the main point here is--generative music doesn't have to have a specific sound if you program it right.
I think one of my confusions that this cleared up is that there is a big difference between generative and algorithmic--while algorithmic is very math-based, generative can be just about anything you want it to be. I suppose algorithmic could, too... but I don't think you can really call anything in reason algorithmic. Just maybe algorithmic emulation. But you could certainly call it generative.
yeah, I think making a deltasleep autogibberish generator would be a truly monumental task. If you wanted it to actually be any good. maybe you could have an AI algorithm read a bunch of nancy drew books, old MAD magazines and country western lyrics and pick out all the best parts.
07/10/07
+
PM |
QUOTE |
PERMALINK |
REPORT
em978
I tend to think that generative music is hard to develop.. in that most artists who use these types of things either don't know that much about math or programming for that matter ... or they get someone else to do the design for them... neither end up with go results. Generative Music can be good but it's more of a question of people putting the time into understanding it.
07/11/07
+
PM |
QUOTE |
PERMALINK |
REPORT
soundhdack
07/11/07
+
PM |
QUOTE |
PERMALINK |
REPORT
eyesnine
i use generative and algorithmic processes in my music, here's my thoughts on the subject:
i don't think a super strong math background is necessary, but a strong background in music theory is. having said that, i take mathematics and computer science in school, so i may be overlooking some concepts i've picked up. music theory is a mathematical system and good familiarity with tones/harmonies and rhythmic structures is essential.
the math really isn't that complicated. music theory uses very basic mathematical structures. if your math gets too complicated your ear won't be able to recognize the structures inherent in the music. this is where i find most academic algorithmic/generative stuff errors. the magic is in your ear, not in the equation. using weather patterns to compose (cage) just seems like a bad idea to me.
i use recursion and repetition a lot. two concepts crucial to chaos theory and fractals. but its really as simple as taking a phrase, halving the tempo and merging it with the original. do this algorithimcally using half/third/quarter tempo manipulations, and place the manipulated phrases appropriately, voila! you have a fractal piece of music. fractals are all about creating complicated patterns by using combinations of basic operations on a seed. it sounds a lot grander than it really is. try making a simple pattern in cubase, copy and paste it all over the place, halve some of the patterns, quarter some of them, double some of them. then transpose the notes for chord changes. no programming necessary.
i guess the most important part of all is the ability to come up with fresh musical systems. not everybody can do this. most people, i find, work below the surface, and while they use mathematical processes, they're not really aware of what they are, and can't put them in the form of a computer algorithm. analyze the music you're already making, tease out the patterns you use and put them in algorithmic form. if you really can't think like this, you'll be stuck just copying other people's ideas, and fooling around with mathematical formulas that have no basis in reality.
as for software, i just use reaktor, and i find its robust enough to realize most of my ideas. other patching languages are suitable (max, pd), as well as text based stuff if you're comfortable with that approach. jsound for java is hot right now because you can embed your applets on a web page. csound is a good starter language if you've never done any music programming.
check out my music if you want to hear what this musical philosophy sounds like. i've been getting good results lately, and i feel like i'm close to another productive period of composition. i haven't made anything in over a month, but i'm gathering some new sounds. hopefully i'll be jamming out in a couple of weeks, and i'll have some new releases to post.
Register / login
|
^
EM411 is Copyright 2001-2008 EM411.com
All rights reserved. | Contact | RSS
|
