[music-dsp] deterministic noise generation

Alexis Glass alexis at andes.ad.kyushu-id.ac.jp
Wed Mar 26 02:52:00 EST 2003


Maybe you could (as Magnus Jonsson else suggested) divide things up by 
bins, but rather than limiting each bin to one sine wave within that 
bin, you could make sure that the contents of each  bin (roughly equal 
in size on a bark scale) are normalized to contain the same amount of 
energy.  I guess that probably wouldn't eliminate the perception of 
clustering or harmonicity, but it might counter it a bit.

Alexis

Svante Stadler wrote:

> I'm generating noise by adding lots of sine waves. My first naive 
> approach was just to generate N frequencies between zero and nyquist 
> from a uniform random distribution. But for N<1000, the sound gets 
> colored due to "clusters" of sines. I need the sines to be more 
> uniformely spaced. But if they are too uniform, their sum will be 
> (pseudo)harmonic, which is not good either.







More information about the music-dsp mailing list