Hello, list. I hope that you're all great.

I just did a pull-request to a library module containing a set of
antialiased nonlinearities. See this:
https://github.com/grame-cncm/faustlibraries/pull/70.

Those include first-order and second-order aliasing suppression based on
Parker et al.:
https://www.dafx.de/paper-archive/2016/dafxpapers/20-DAFx-16_paper_41-PN.pdf
.

Julian et al.: if you're reading this, thanks again for the great work.

Initially, I tried implementing the algorithms from Bilbao et al.:
https://aaltodoc.aalto.fi/bitstream/handle/123456789/30723/article5.pdf?sequence=8&isAllowed=y
.

This is an extension of Parker et al. work covering also third-order
antialiasing. It would have been great to have those but I never managed to
make them work, not even the second-order functions. From what I could
tell, the second-order approach from Parker et al. is more robust, whereas
the handling of ill-conditioned cases in Bilbao et al. may be trickier.

Or most likely I did something wrong, which is why I'm handing over to
anybody interested what I already started, hoping that you can achieve
better results, at least with the second-order functions: I have spent more
time than I had on this and I'm happy with the second-order functions in my
PR. However, if we get the second-order functions from Bilbao et al. to
work, we might be close to having the third-order ones too, which would be
excellent.

See the file attached. That's the skeleton for a second-order tanh() using
an approximation of the dilog function, but it seems that the only way to
actually have it working efficiently is to use a look-up table. Anyway, I
tried that with simpler nonlinearities and it didn't work.

If you want to try, once you think that the ADAA2 generalised function is
correct, just add first and second antiderivatives for F1 and F2.

Ciao.

-- 
Dr Dario Sanfilippo
http://dariosanfilippo.com

Attachment: ADAA2_Bilbao.dsp
Description: Binary data

_______________________________________________
Faudiostream-users mailing list
Faudiostream-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/faudiostream-users

Reply via email to