Designing Sound in SuperCollider/Creaking door

Fig 32.3: formants for a wooden door

(
~woodfilter = { |input|
	var freqs, rqs, output;
	// Note: these freqs are as given in the diagram:
	freqs = [62.5, 125, 250, 395, 560, 790];
	// The Q values given in the diagram (we take reciprocal, since that's what BPF unit wants)
	rqs   = 1 / [1, 1, 2, 2, 3, 3];
	// in the text, andrew says that the freqs follow these ratios, 
	// which give a very different set of freqs...:
	// freqs = 125 * [0.5, 1, 1.58, 2.24, 2.92, 2, 2.55, 3.16];
	
	//Now let's apply the parallel bandpass filters, plus mix in a bit of the original:
	output = BPF.ar(input, freqs, rqs).sum + (input*0.2);

};

// Now let's use this function in something - some dust impulses tap-tap-tapping on the door:
x = {
	Pan2.ar(~woodfilter.value(  LPF.ar(Dust.ar(10), 10000)  ))
}.play;
// Doesn't sound much like a door? Compare it to the raw tapping sound (edit the above).
)

Fig 32.4: stick-slip motion in response to applied force

(
~stickslip = { |force|
	var inMotion, slipEvents, forceBuildup, evtAmp, evtDecayTime, evts;
	force = force.lag(0.1); // smoothing to get rid of volatile control changes
	
	inMotion = force > 0.1; // static friction: nothing at all below a certain force
	
	// slip events are generated at random with freqency proportional to force.
	// I originally used Dust to generate random events at a defined frequency, but
	// that lacks the slight "pitched" sound of the creaky door. Here we use Impulse
	// to generate a frequency, but we add some noise to its frequency to try and 
	// avoid it getting too perfectly regular.
	slipEvents = inMotion * Impulse.ar(force.linlin(0.1, 1, 1, 1/0.003) * LFDNoise1.ar(50).squared.linexp(-1,1, 0.5, 2).poll);
	
	forceBuildup = Phasor.ar(slipEvents, 10 * SampleDur.ir, 0, inf).min(1);
	
	// Whenever a slip event happens we use Latch to capture the amount of
	// force that had built up.
	evtAmp = Latch.ar(Delay1.ar(forceBuildup.sqrt), slipEvents);
	evtDecayTime = evtAmp.sqrt;
	// The book applies square-root functions to shape the dynamic range of the events.
	// Remember that square-root is computationally intensive, so for efficient 
	// generation we might want to change it to (e.g.) a pre-calculated envelope.
	
	// Now we generate the events
	evts = EnvGen.ar(Env.perc(0.001, 1), slipEvents, evtAmp, 0, evtDecayTime * 0.01);
};
// Let's plot 4 seconds worth, with steadily increasing force.
// Events should appear more frequent but less violent as the plot progresses.
{~stickslip.value(Line.kr(0,1,4))}.plot(4);
)

Fig 32.5: parallel delays to simulate rectangular door frame

(
~squarepanel = { |input|
	var times, filt;
	// times in milliseconds, converted to seconds:
	times = [4.52, 5.06, 6.27, 8, 5.48, 7.14, 10.12, 16] * 0.001;
	filt = DelayC.ar(input, times, times).mean;
	filt = HPF.ar(filt, 125);
	filt * 4
};
)

Fig 32.6: putting it all together

The following re-uses the functions defined further up this page in combination, to create a door which you can push yourself!

Put the mouse over to the left of the screen before you run this code. Then move the mouse to the right and to the left, to control the amount of force on the door.

x = {~squarepanel.value(~woodfilter.value(~stickslip.value(MouseX.kr(0,1))))}.play