«003_054» by evilshadowtwin

on 02 Aug'13 21:48 in unnamable
1
SynthDef('003_054', {|out=0,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y|  Out.ar(out, Pan2.ar(Normalizer.ar((  CuspL.ar(e, e,  SawDPW.ar(x,  SyncSaw.ar( PMOsc.ar(a, g, k, a, x, f ), j, b, p ), v, i ),  SyncSaw.ar(u,  HenonL.ar(i, m, c, t, l, r, f ),  GbmanL.ar(m, x, s, f, y ), n ), o, b ))*( PMOsc.ar( LFSaw.ar(l, b, p, q ), t, x,  LFTri.ar(l,  LFPulse.ar( Spring.ar(d, x, f ), v, i, k, h ),  LFPulse.ar(l, n, h, n, v ), r ), k, q ))*( Formant.ar(d, x,  LFDNoise1.ar(l,  LFCub.ar(k, l, s, v ),  LFClipNoise.ar(f, m, p ) ),  SinOsc.ar(g,  CML0.ar(g, l, i, c, i, f ),  LFPulse.ar(l, f, c, b, s ),  LFNoise1.ar(l, x, v ) ),  LPF.ar( SyncSaw.ar(m, q, u, r ),  BrownNoise.ar(t, h ), f, n ) ))*( QuadL.ar(x,  LFNoise1.ar( CombL.ar(e, 1, x, w, d, c ), u,  Blip.ar(b, q, m, b ) ),  LFPar.ar(h,  LFClipNoise.ar(u, v, v ),  LFClipNoise.ar(n, d, j ), m ), k,  SinOsc.ar(x, l, o, l ), m,  HenonL.ar(p, w, h, p, a, j, w ) ) )))) }).play(args: #[ a, 0.45186975651959, b, 4.3496208121178, c, -3.5050481728403, d, 0.21674529882104, e, 3.6302846277664, f, 2.0252093014502, g, -0.50985145568848, h, 4.7362252591183, i, -1.1713491709346, j, 11.646696339178, k, 0.66817343235016, l, 0.11935562425443, m, 1.350574016571, n, -1.7757006468997, o, 2.0951379285004, p, 3.5410180366402, q, 0.39763283729553, r, 0.93207371234894, s, 6.2958073430114, t, 6.1286854287138, u, 4.8556221894114, v, 4.8556221894114, w, 1.7277424641031, x, 4.8556221894114, y, 2.9251942534878 ])
raw 1451 chars (focus & ctrl+a+c to copy)
reception
comments
vividsnow user 08 Aug'13 12:01

got "class not defined" - CML0

p.s. why single line? )

evilshadowtwin user 15 Aug'13 18:39

CML0 is part of Yota Morimoto's nlugens: https://github.com/yotamorimoto/nl4sc

code generated by an evolutionary algorithm, not for human consumption.

vividsnow user 16 Aug'13 08:03

thanks for pointing! what did you choose as fitness function?

evilshadowtwin user 16 Aug'13 11:11

the fitness is determined by comparing machine listening data of the generated functions to those of an existing sound

vividsnow user 19 Aug'13 15:17

interesting! what did you use as a metric of data? ..power spectrum / whole record / something else?

evilshadowtwin user 29 Aug'13 09:20

mean + stddev of 20 MFCC, SpecFlatness, SpecCentroid, Amplitude. any synthdef producing even a single NaN or inf must DIE...