Main Archive Specials Wiki | FAQ Links Submit Forum

Delay time calculation for reverberation

References : Posted by Andy Mucho

Notes :
This is from some notes I had scribbled down from a while back on
automatically calculating diffuse delays. Given an intial delay line gain
and time, calculate the times and feedback gain for numlines delay lines..

Code :
int numlines = 8;
float t1 = 50.0; // d0 time
float g1 = 0.75; // d0 gain
float rev = -3*t1 / log10 (g1);

for (int n = 0; n < numlines; ++n)
float dt = t1 / pow (2, (float (n) / numlines));
float g = pow (10, -((3*dt) / rev));
printf ("d%d t=%.3f g=%.3f\n", n, dt, g);

The above with t1=50.0 and g1=0.75 yields:

d0 t=50.000 g=0.750
d1 t=45.850 g=0.768
d2 t=42.045 g=0.785
d3 t=38.555 g=0.801
d4 t=35.355 g=0.816
d5 t=32.421 g=0.830
d6 t=29.730 g=0.843
d7 t=27.263 g=0.855

To go more diffuse, chuck in dual feedback paths with a one cycle delay
effectively creating a phase-shifter in the feedback path, then things get
more exciting.. Though what the optimum phase shifts would be I couldn't
tell you right now..


Added on : 11/09/08 by bob[ AT ]yahoob[ DOT ]com
Comment :
Hello, when you say 'dual feedback paths with one cycle delay' do you mean dual as in stereo? And one cycle means one sample?
My experiments have massive energy build up of energy unless the feedback was something like * 0.01, but there was still ringing in the sound. Can you hint a little more about what you mean?

Add your own comment
Comments are displayed in fixed width, no HTML code allowed!


Are you human?

Site created and maintained by Bram
Graphic design by line.out | Server sponsered by fxpansion