Moin Phil, Thanks for the comments.
Regarding the Voss algorithm, the midrange peaks show indeed the strong aliased terms of zero-order hold. Regading the pk3 algorithm, I would love to see an implementation faster than mine, but even with AVX SIMD, it depends largely on the voodoo performed for generating the white noise input. If there would be a faster SIMD implementation of pk3, my new algorithm has still the advantage that it can go very deep, only limited by the number of bits in your word/mantissa, with almost no performance loss. Think 50 octaves, fs=44100, then wait about 800 years for the first update of the lowest octave. For audio purpose this is nonsense, but there might be application where you do want to go very low. Stefan On 07 May 2014, at 18:31 , Phil Burk <philb...@mobileer.com> wrote: > Hello Stefan, > > > http://stenzel.waldorfmusic.de/post/pink/ > > This is really interesting. I love being able to turn on or off various > octaves and hear the effect using your JavaScript implementation. > > Adding the FIR extends the high end nicely. I love your trick of precomputing > the FIR by taking advantage of single bit inputs. And you combine the > shifting of the bits in the LFSR with the delay of the input values to the > FIR. Nice. > > I notice that the error curve for the Voss algorithm has lots of peaks in the > midrange compared to your algorithm. Is that mostly because you do linear > interpolation instead of zero-order hold? > > I also notice that the pk3 algorithm has low error and is almost as good as > yours. It is the sum of 6 first order filters, using 14 multiplies. Given the > speed of multiplies on modern processors, is it possible that the pk3 > algorithm is faster than yours? The filters are independent so it would work > well on a SIMD architecture. > > Phil Burk > > On 5/7/14, 1:20 AM, Stefan Stenzel wrote: >> Quick and quite accurate pink noise generator, maybe useful for someone: >> http://stenzel.waldorfmusic.de/post/pink/ >> >> Stefan >> -- >> dupswapdrop -- the music-dsp mailing list and website: >> subscription info, FAQ, source code archive, list archive, book reviews, dsp >> links >> http://music.columbia.edu/cmc/music-dsp >> http://music.columbia.edu/mailman/listinfo/music-dsp >> > -- > dupswapdrop -- the music-dsp mailing list and website: > subscription info, FAQ, source code archive, list archive, book reviews, dsp > links > http://music.columbia.edu/cmc/music-dsp > http://music.columbia.edu/mailman/listinfo/music-dsp -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp
