On Wed, 4 Jan 2006 11:41:07 -0500 Marc Lavallée marc@hacklava.net wrote: <snip>
Jamie, I checked your external (http://www.puredata.org/Members/jb/fct/view) I see that you are already releasing everything with the original ACM license. I think your port of the original FORTRAN routine (http://www.netlib.org/toms/749) does not belong to the ACM.
Well, I saw that as the most obvious thing to do. The problem is that I would like to move all of my externals into the PD CVS, and I understand that in order to do this they need to be licensed under the GPL. Hence my question.
The original code is definitely licensed under the ACM license. The TOMS link you provide above is for Transactions on Mathematical Software, which is an ACM publication.
I'm curious about the Fast Cosine Transform; is it faster than a DCT or a FFT? Why not using the fftw library?
Believe it or not, when I wrote the fct~/ifct~, I didn't realise that it was possible compute the DCT from an FFT! I wanted to do some MFCC computations in PD, so I found a Cosine transform algorithm that I thought would be relatively fast and quick to implement. I don't think it even performs as well as PD's Meyer FFT, but this is possibly due to my crude programming skills rather than the algorithm itself. I learned a lot doing it though.
Jamie