|
|
Terence Hill
from Germany on 2009-11-04 16:41 [#02342238]
Points: 2070 Status: Lurker
|
|
hi been playing around with genetic algorithms a bit so here's the first test result, i call it "LAZYBARF". it's generatively real time sequenced with fancy custom software, that's why all the girls love it too.
|
|
JivverDicker
from my house on 2009-11-04 16:58 [#02342239]
Points: 12102 Status: Regular | Followup to Terence Hill: #02342238
|
|
It's going to be No.1 all over the world!
|
|
w M w
from London (United Kingdom) on 2009-11-04 17:51 [#02342258]
Points: 21454 Status: Regular
|
|
I find music based on code interesting. As far as enjoyment of listening to it, it was ok but nothing that special I guess imo. Some argue that you can ignore jivverdicker because he's a fag's supernumerary nipple.
|
|
Terence Hill
from Germany on 2009-11-05 07:41 [#02342332]
Points: 2070 Status: Lurker | Followup to w M w: #02342258
|
|
thanks and yeah i'm really not happy with it myself, had to spam it either way. this whole generative stuff proves to be more difficult in detail than imagined, actually i wanna trash mah computer now and plant tomatoes instead.
|
|
mohamed
from the turtle business on 2009-11-05 08:14 [#02342334]
Points: 31231 Status: Lurker | Followup to Terence Hill: #02342332 | Show recordbag
|
|
tomatina 2009
|
|
Terence Hill
from Germany on 2009-11-05 08:39 [#02342340]
Points: 2070 Status: Lurker | Followup to mohamed: #02342334
|
|
this is excellent and makes much more sense than crunching numbers, thanks :D
LAZY_TITLE
|
|
Tractern
from Brighton (United Kingdom) on 2009-11-05 10:08 [#02342347]
Points: 4210 Status: Regular | Show recordbag
|
|
Interesting.
BTW, your avy is a jerkface.
|
|
Terence Hill
from Germany on 2009-11-05 10:14 [#02342348]
Points: 2070 Status: Lurker
|
|
LAZYBARF V2 might be better but maybe not. It's shorter sequences and i left out scale modulation.
|
|
khrimson
from the fridge on 2009-11-05 12:30 [#02342370]
Points: 1757 Status: Regular
|
|
haha, grande saladin from civIV
|
|
khrimson
from the fridge on 2009-11-05 12:32 [#02342373]
Points: 1757 Status: Regular
|
|
on topic, the track sounds cool, reminds me of this: LAZY_TITLE
|
|
Terence Hill
from Germany on 2009-11-05 16:55 [#02342418]
Points: 2070 Status: Lurker | Followup to khrimson: #02342373
|
|
/lol
|
|
Terence Hill
from Germany on 2009-11-06 08:46 [#02342543]
Points: 2070 Status: Lurker
|
|
due to astounding public demand, here's more genetic LAZYBARF.
|
|
khrimson
from the fridge on 2009-11-06 12:17 [#02342577]
Points: 1757 Status: Regular
|
|
I like v2, care to share/explain more about this? also, it is genetic sequencing I guess, so you input a patch and a basic pattern to evolve from?
|
|
Terence Hill
from Germany on 2009-11-06 15:29 [#02342600]
Points: 2070 Status: Lurker | Followup to khrimson: #02342577
|
|
so what it is is 2 instances of a monophonic sequencer with variable step length (steps are spread across different channels though, i'm using 12 channels in Ableton)... at the beginning it creates one semi-random target sequence and a population of 2000 random sequences, then there's fitness evaluation/selection/mutation/mating of the fittest. what you hear is always the current fittest sequence, i.e. the one which is closest to target... so after a while when the population has evolved really close to the target sequence, a new target is generated and so on.
the genome, i.e. the sequence parameters, is mainly note values/step lengths/channels/velocity in this case... this is just first baby steps in GA for me, like mentioned above i'm not happy with it so far. I want to have an evolving eco system where different "species" (phrases, patterns, ...) compete for survival, but this will be really complex to build.... and several magnitudes more difficult to make it sound nice, that's actually the most daunting part - which parameters are mapped to what and what values do they have :D
|
|
khrimson
from the fridge on 2009-11-06 16:51 [#02342619]
Points: 1757 Status: Regular
|
|
so it does compute another generation each measure? If not I don't get it
|
|
Terence Hill
from Germany on 2009-11-06 17:15 [#02342621]
Points: 2070 Status: Lurker
|
|
it does compute ~50 generations per second, only the sequencer is updated with new data after each measure. i've built on top of this example...
so yeah it's very basic GA - the goal is known from the beginning, then there's evolution from random numbers towards this goal. more interesting would be fitness evaluation based on some kind of arbitrary, unknown, "aesthetic survivability" but ehhhh, try to teach a machine what's aesthetically pleasing...
|
|
JivverDicker
from my house on 2009-11-06 17:21 [#02342623]
Points: 12102 Status: Regular
|
|
It's a trivial game on the pseudo academic level but it sounds shit. What are you trying to say?
|
|
JivverDicker
from my house on 2009-11-06 17:24 [#02342624]
Points: 12102 Status: Regular | Followup to Terence Hill: #02342621
|
|
It's not hard if you pick some great music and apply that to well thought out algorithms. You still need an idiot doing stuff to make it whole.
|
|
khrimson
from the fridge on 2009-11-06 17:25 [#02342625]
Points: 1757 Status: Regular | Followup to Terence Hill: #02342621
|
|
that's was my quirk, if you know the target and you try to best fit you get something that is 99,9% similar to the target, no? yet there are variations in that last piece so well, it's nice
|
|
JivverDicker
from my house on 2009-11-06 18:06 [#02342630]
Points: 12102 Status: Regular | Followup to khrimson: #02342625
|
|
Haha! What crap logic.
|
|
Terence Hill
from Germany on 2009-11-06 19:03 [#02342631]
Points: 2070 Status: Lurker | Followup to JivverDicker: #02342623
|
|
a fag's supernumerary nipple
|
|
Messageboard index
|