noise generation

James McCartney asynth at
Fri Sep 3 02:09:11 EDT 1999

At 9:14 PM -0600 9/2/99, Robin Whittle wrote:
>James McCartney wrote, in part, what may henceforth be cited as the
>"Orfanidis-Gardner-McCartney 1/f noise algorithm":
>> Now the problem with the above is that on some samples you have to
>> add up a lot more random values than others. This can also cause
>> large discontinuities in the wave when lots of the values
>> change at once.
>> By rearranging the order that the random generators change the
>> load can be made even, the operation can be made more efficient
>> and the deviation from any one sample to another is bounded by a
>> constant value every sample. Here's the pattern, a tree structure:
>> x x x x x x x x x x x x x x x x
>>  x   x   x   x   x   x   x   x
>>    x       x       x       x
>>        x               x
>>                x
>> ( . . . )
>This looks extremely elegant.  However, I understand that the higher
>frequencies need to be at a lower level than the lower frequencies, so
>those step-wise jumps in the least-often sampled noise sources will be
>large by comparison with the steps of the more frequently sampled

No. All are at equal amplitude. Remember that 1/f noise has equal
power per octave. Each octave is twice as wide in cps at the next lower.

   --- james mccartney     james at
If you have a PowerMac check out SuperCollider2, a real time synth program:

dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links

More information about the music-dsp mailing list