supernova: fix for small audio vector sizes

[supercollider.git] / examples / pieces / microhelix_study.scd

(
//a study for recreating key (sound) aspects of data.microhelix by Ryoji Ikeda
//Batuhan Bozkurt 2009 http://www.earslap.com
{
        //rhythm logic: very crude, times are 1/8's mixed with 1/4's (equal probability) plus 1/16. 
        //when 1/16 is chosen, we make sure they come in doubles so it sounds rhytmically nicer.
        //we create 2 instances of these trigs so glitching sound will be independent between speakers.
        //we use TDuty with Demand UGens to create this kind of sophisticated triggers (easier and more 
        //flexible dealing with pure noise generators).
        var ctrigs = 
        {
                TDuty.ar
                (
                        Dxrand(((1/8!8) ++ (1/4!8) ++ [Dseq([1/16, 1/16], Drand([1, 2], inf))]) * 1.25, inf), //see above logic
                        0, 
                        Dwhite(0.5, 1, inf) //trigger values are between 0.5, 1
                )
        }!2; 
        
        //the glitch/clicking noises: basically enveloped PinkNoise through bandpass filter with high rq (low q)
        //here I'm modulating the decay time of the noise envelope with a LFNoise1.
        //there are many diferent ways of approaching this, needs experimentation,
        //dynamic center freq, rq etc. can be employed for example. This sounded nice to me, however.
        
        var clicks = 
                BPF.ar
                (
                        PinkNoise.ar(Decay.ar(ctrigs, 0.001 * LFNoise1.ar(4).abs)), //decay time is modulated
                        15000, 
                        0.9, 
                        25 * LFNoise1.ar(8).range(0, 1) //extreme amplification of glitches. 
                ).fold(-1, 1); //folding them back into [-1, 1] range (foldback distortion). I think it sounds nice.
        
        //i don't really know what to call this sound, so its names snd1. 2 sine oscillators tuned by ear, with filtered
        //background noise. nothing fancy. we will random-pan and envelope this later on.
        var snd1 = 
                LPF.ar
                (
                        SinOsc.ar(44.midicps, 0, 0.5) + SinOsc.ar(90.midicps, 0, 0.6), 
                        32.midicps * 2
                ) + HPF.ar(LPF.ar(WhiteNoise.ar(0.008), 12000), 2400);
        
        //the shaker-like sound in the background. in the original piece, the rhytmical logic for this sound
        //is a lot more different, I'm using the same ctrigs here, for the sake of simplicity (and I'm a bit lazy).
        //its basically an enveloped WhiteNoise through a bandpass filter with a random freq 
        var hiNoise = 
                BPF.ar
                (
                        WhiteNoise.ar(Decay2.ar(ctrigs * LFNoise1.ar(8, 0.5, 0.5), 0.02, 0.1) * 0.05), 
                        TRand.ar(12000, 15000, ctrigs), 
                        0.9
                );
        
        //the bass/kicky sound needs a lot of tuning (sound and envelope, some best be left to post processing maybe)
        //2 sine oscillators, one has a low freq nad the other is an upper harmonic (quieter)
        //the original piece uses a completely different (and very better) rhytmical logic for this sound.
        //I'm using our ctrigs impulse generator here (first channel) for the sake of clarity.
        //but since I don't want every trigger to activate this, I'm multiplying the trigger with
        //a noise generator (rounded, so its -1, 0, or 1). so not all trigs pass through.
        
        //the sine oscillator uses a trick I've learned from Josh Parmenter. the freq arguments of the
        //sine oscillators are set to zero, and they are driven by the phase argument with a sweep.
        //this makes the oscillator phase resettable which is very useful for having uniform attacks
        //for low frequency kick-like sounds. or else, the envelope sounding the oscillators can kick
        //in at any phase and the attack of the kicks will vary (in a bad way). here I reset the phase  
        //of the sine at each trigger. the speed of the phase driver ([52.8, 740]) determines the 
        //frequency of the oscillator here. I'm adding a phase of pi/3 to it so the phase resets to 
        //pi/3 instead of zero which produces an additional glitch. this is all about glitch after all right?
        
        var trigMod = LFNoise0.ar(8).round;
        
        var bass = 
                SinOsc.ar(0, (Sweep.ar(ctrigs[0] * trigMod, 2pi * [52.8, 740]) + (pi/3)).wrap(-pi, pi), [2, 0.05]).mean.tanh * 
                EnvGen.ar(Env([0, 0.5, 0.4, 0], [0, 0.2, 0.01], -5), (ctrigs[0] * trigMod).abs)!2; 
        
        //panning and enveloping of snd1. I don't want to use every trigger for the envelope of this so
        //i again multiply it with a random generator. not all trigs pass through. and some random panning.
        snd1 = 
                Pan2.ar
                (
                        snd1 * EnvGen.ar(Env([0, 1, 0.6, 0], [0.0001, 0.4, 0.01]), ctrigs * LFNoise0.ar(8)), 
                        TRand.ar(-1, 1, ctrigs)
                );
        
        //i'm using limiter here because i'm lazy. one needs to get the
        //gain scheduling right for a better sound. this especially kills the bass
        //sometimes. I'm also boosting the freqs around 15000Hz.
        //delete some sounds from the input of MidEQ to hear the sounds
        //(glitches, bass etc) in isolation.
        Limiter.ar(MidEQ.ar(clicks + snd1 + hiNoise + bass, 14000, 0.7, 8));
}.play;
)