"True" pink noise is likewise impossible: if it's equal energy per octave, then since there is no "lowest octave," any energy in any octave implies infinity at 0hz. The fexpr~ approach I gave accumulates, but imagines a "container" with "walls" at +/- 1.0 where whatever was doing the random walk "bounces" off the wall. It should have a color within auditory range similar to unrestricted integration. A lot depends on what you want to do with the noise (it's not necessarily an audio source).
Convolution is a pretty good way to go. I've seen a number of filters that try to distribute phase randomly across the frequency magnitude curve. Also remember that the 6dB dropoff is with respect to log frequency, whereas FFT is linear in frequency.
On Sat, Nov 14, 2015 at 1:35 PM, Martin Peach chakekatzil@gmail.com wrote:
There's [rojo~] in mrpeach, if by brown noise you mean red noise. (brownian noise?) [rojo~] tends to grow out of bounds, which is a problem with correlated noise: if you want long term correlations you have to be ready for the noise to get very loud. So in practice there is no red noise possible, it's all high pass filtered. To get any arbitrary shape of noise it can be done most efficiently by convolving a frequency curve table with white noise. The table would have the same length as the number of bins in the FFT.
Martin
On Fri, Nov 13, 2015 at 10:35 PM, Alexandre Torres Porres < porres@gmail.com> wrote:
hi, wondering how to implement brown noise in Pd (vanilla or extended) - moreover, I'd like and if there's an external that does it.
same about grey noise
thanks
alex
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list