Designing Sound in SuperCollider/Sirens

From Wikibooks, open books for an open world
Jump to: navigation, search

28: US-style police siren[edit]

We'll use the internal server to make sure we can use the oscilloscope etc.:

Server.default = s = Server.internal;
s.boot;

Fig 28.6: imitation of capacitor charge/discharge[edit]

In SuperCollider we can provide an almost-direct physical model: the LFPulse represents a "raw" on/off signal before smoothing by capacitors, and the "lagud" provides exponential smoothing, with the handy feature of allowing different time-periods for the "on" and "off" convergence. This one-line example will plot the curve for you:

{LFPulse.ar(1, 0.99, 0.4).lagud(0.3, 0.7)}.plot(2)

The siren[edit]

Now let's use this technique both for the pitch curve, and for waveform synthesis.

(
SynthDef(\dsaf_horn1, { |rate=0.1|
	var freq = LFPulse.kr(rate, 0.99, 0.4).lagud(0.4 / rate, 0.6 / rate) * 800 + 300;
	var son  = LFPulse.ar(freq, 0.99, 0.2).lagud(0.4 / freq, 0.6 / freq) * 2 - 1;
 
	// This filtering is a simple approximation of the plastic horn acoustics:
	son = BPF.ar(son.clip2(0.2), 1500, 1/4) * 4;
 
	// delay and reverb, to simulate the environment in which we hear the siren
	son = son + DelayC.ar(son, 0.1, 0.1, 0.3);
	son = son + FreeVerb.ar(son);
 
	Out.ar(0, Pan2.ar(son * 0.4));
}).add;
)
 
x = Synth(\dsaf_horn1);
 
s.scope
 
// Choose a rate
x.set(\rate, 3);
x.set(\rate, 0.1);

Exercise: instead of using the lagged-pulse implementation of the waveform, do as the book says and try using a simple triangle oscillator (LFTri) - this loses the physical-modelling "realism" of the authentic circuit but will be more efficient, and reasonably similar. How much of an effect on the tone quality do you get?