Hi ~Kyle,
But is the algorithm only supposed to generate 1/2 of the hfa array? I "doodled" in it, just to play with the convolution, and now the agorithm is only writing to the first 512 locations of the array. Is it a bug?
It is wanted to be like that; actually in the original algo
there are arrays (of noise and hfa coeff) of 2n elements: n
values are filled, the other are padded with zeros.
Then real fft is done on the first n and convolution on 2n
elements.. so mine is an adaptation of this. When I looked
at the spectrum (in audacity) the noise seems to behave
fine; my concern is the strange sensation
in hearing the noise, like a small "feedback" even on a
single channel of output for high alpha..
Alberto
On 8/19/06, alberto.zin@poste.it alberto.zin@poste.it wrote:
Hi,
attached is an abstraction for generation of "colored" noise (http://en.wikipedia.org/wiki/Colors_of_noise) using a relatively new method (by J.N. Kasdin, 1995) that uses frequency domain multiplication. White, pink, brown noise (and intermediate flavors) can be obtained by varying a single parameter (the slope of the spectrum in a linear log(PSD)-log(f) plot). I searched a bit in the archives and it seems to me that in for noise generation in PD we actually have noise~ (uniform white noise), pink~ (from zexy and cyclone) and a variety of filters, but no direct generation of 1/f^a noise (am I right?).
Inside the archive there is a short readme, look at it first. To me the patch sounds a little "tubby", let me know if you have the same feeling.
Any comment, suggestion for improvements is welcome.
The abstraction lives in http://www.puredata.org/Members/AlbertoZ
All the best,
Alberto Zin
PD-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
hello.
i can't explain this feedback. perhaps it is due to some iteration
when reading some table. anyway, there is a problem in the
abstraction : noise~, as you say, is not a gaussian noise source. it
is a uniform white noise source.
to have a gaussian noise source in pd, i suspect a pd external should
be writen. Karl-L. Noell and Helmut Weber wrote a library that
implements the Marsaglia uniform random numbers generator. Inside,
there is a routine to generate gaussian numbers. I used it to write
some externals for maxmsp. the sources, that are public, could be
easily adapted to pd ...
http://www.sonoscop.net/jmb/pub/gaussmt~.zip (gaussian number
generator using mersenne twister uniform random number generator)
http://www.sonoscop.net/jmb/pub/gaussm~.zip (gaussian number
generator using tt80 uniform random number generator)
http://www.sonoscop.net/jmb/pub/gaussr250.zip (gaussian number
generator using r250 uniform radom number generator)
http://www.sonoscop.net/jmb/pub/gauss~.sit (gaussian number generator
using Marsaglia uniform random number generator)
http://www.sonoscop.net/jmb/pub/rojo~.zip (red noise generator using
Marsaglia uniform random number generator )
http://www.sonoscop.net/jmb/pub/ruidist.zip (red noise, 1/f^alpha
noise, random number generator and statistical distributions)
http://www.sonoscop.net/jmb/pub/ruidos~.zip (marsaglia, tt80 and
mersenne twister uniform random number generators)
jmb
El 20/08/2006, a las 11:37, alberto..zin@@poste..it escribió:
Hi ~Kyle,
But is the algorithm only supposed to generate 1/2 of the hfa array? I "doodled" in it, just to play with the convolution, and now the agorithm is only writing to the first 512 locations of the array. Is it a bug?
It is wanted to be like that; actually in the original algo there are arrays (of noise and hfa coeff) of 2n elements: n values are filled, the other are padded with zeros. Then real fft is done on the first n and convolution on 2n elements.. so mine is an adaptation of this. When I looked at the spectrum (in audacity) the noise seems to behave fine; my concern is the strange sensation in hearing the noise, like a small "feedback" even on a single channel of output for high alpha..
Alberto
On 8/19/06, alberto.zin@poste.it alberto.zin@poste.it wrote:
Hi,
attached is an abstraction for generation of "colored" noise (http://en.wikipedia.org/wiki/Colors_of_noise) using a relatively new method (by J.N. Kasdin, 1995) that uses frequency domain multiplication. White, pink, brown noise (and intermediate flavors) can be obtained by varying a single parameter (the slope of the spectrum in a linear log(PSD)-log(f) plot). I searched a bit in the archives and it seems to me that in for noise generation in PD we actually have noise~ (uniform white noise), pink~ (from zexy and cyclone) and a variety of filters, but no direct generation of 1/f^a noise (am I right?).
Inside the archive there is a short readme, look at it first. To me the patch sounds a little "tubby", let me know if you have the same feeling.
Any comment, suggestion for improvements is welcome.
The abstraction lives in http://www.puredata.org/Members/AlbertoZ
All the best,
Alberto Zin
PD-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
jose manuel berenguer jmbeal@telefonica.net +34932857046 +34696538403. http:// www.sonoscop.net/jmb/ jmberenguer@sonoscop.net +34933064128. http://www.sonoscop.net/ caos->sonoscop. cccb. montalegre, 5. 08001 barcelona spain
-
alberto.zin＠poste.it -
jose manuel berenguer