Hi guys,
I’d like to share this tape delay implementation I came up with the last few days.
I’ve tried to put into SC my theoretical knowledge of the actual old-days tape delay machines.
This is the implementation:
(
SynthDef(\tapedly, {
|in=0, out=0,fb=0.6, fc=5000, dlytime=0.125, wet=0.5, pan=0.0|
var local, sig, delayed, phi, phR, phP, freq;
var tdbh; // time delay between heads (seconds)
local = LocalBuf(SampleRate.ir * 3, 1);
local.clear; // clear the buffer
tdbh = 0.125;
phi = (2 * tdbh) / BufDur.ir(local);
freq = 1.0 / BufDur.ir(local); // frequency to playback the buffer at rate 1
freq = (freq * tdbh) / VarLag.kr(dlytime, 2, warp:\linear);
phR = LFSaw.ar( freq, phi, 0.5, 0.5)*BufFrames.kr( local );
phP = LFSaw.ar( freq, 0.0, 0.5, 0.5)*BufFrames.kr( local );
sig = Mix.ar(In.ar(in,2));
delayed = BufRd.ar(1, local, phP, loop:1);
delayed = delayed.tanh;
BufWr.ar( LPF.ar( (sig + delayed), fc, mul:fb), local, phR, loop:1);
pan = LFNoise1.ar(5);
Out.ar(out, Pan2.ar(delayed*wet, pan));
}).add;
)
And you can test it using these line of code:
// instantiate the delay and eventually set some of its parameters
d = Synth(\tapedly, [\in, 0, \dlytime, 0.333, \fb, 0.5, \wet, 1.0]);
d.set(\dlytime, 0.25, \fb, 0.8,\wet, 1.0,\fc, 5000)
// define a test synth
(
SynthDef(\pulse, {
|freq=440, fb=0.5, amp=1.0, atk=0.0, rel=0.1, out=0, pan=0.0|
var env = EnvGen.ar(Env.perc(atk, rel), 1, doneAction:2);
var sig = SinOscFB.ar(freq, fb) * env * amp;
Out.ar(out, Pan2.ar(sig, pan));
}).add;
)
// and use it via a Pbindef
(
Pbindef(\testPulses,
\instrument, \pulse,
\scale, Scale.minor,
\octave, Prand([4,5,6],inf),
\degree, Prand([-1,0,2], inf),
\fb, Pexprand(0.1, 0.7, inf),
\atk, 0.01,
\rel, Prand([0.125, 0.25, 0.5], inf),
\dur, Prand([0.25,0.5],inf),
\amp, 0.25,
\out, 0,
\pan, 0.0,
\addAction, Pwrand([0,1],[3, 10].normalizeSum,inf)
).quant_(1).play;
)
If you want you can easily control the synth parameters via MIDI controls (which is more funny to me 
):
MIDIClient.init;
MIDIIn.connectAll;
(
~ctrl_1.free;
~ctrl_1 = MIDIFunc.cc({
arg ...args;
args.postln;
d.set(\dlytime,args[0].linlin(0, 127, 0.125, 1.0) );
},1);
)
(
~ctrl_2.free;
~ctrl_2 = MIDIFunc.cc({
arg ...args;
args.postln;
d.set(\fb,args[0].linlin(0, 127, 0.0, 0.99) );
},2);
)
(
~ctrl_3.free;
~ctrl_3 = MIDIFunc.cc({
arg ...args;
args.postln;
d.set(\fc,args[0].linexp(0, 127, 100.0, 5000) );
},3);
)
I’ve preferred to use LFSaw instead of Phasor implementing the way to read and write on the internal buffer (which represent the tape loop). I’ve also keep a “physical” distance between the recorder an player heads (in code it is expressed in terms of time instead of space by the “tdbh” variable a.k.a. “time delay between heads”).
I’m quite happy with this but I’m pretty sure this could be improved in terms of implementation, accuracy and efficiency.
What do you think about it? Let me know,
Thank you so much
na