> I think the main problem is when creating digital waveguide filters.  The
> vdelay and deltapi opcodes use linear interpolation.  This becomes a problem
> at high frequencies.  44.1K/5K = 8 samples which is not sufficient for
> accurate tuning.  An improved version should be written which use sinc
> interpolation which is commonly found in fractional digital waveguide
> filters.  If someone has deep understanding of such things please write us a
> fractional waveguide filter opcode.

While i don't have deep understanding of the topic, you may be interested
in the sinc interpolation code I wrote a couple years ago.  It's an app
that implements the digital analogon to analog tape "scrubbing", forward
and backwards.  While the computation is heftier than simple linear 
interpolation or even cubic interpolation, it has the great advantage
that computing costs increase only linerarly with quality.  In fact, on 
many modern machines it tends to be faster even than cubic interpolation.
I think SGI implemented the algorithm now in their new digital media conversion library, for exactly that reason.

Check out

  http://ccrma-www.stanford.edu/~tkunze/sw/index.html#scrub

Comes with sources, of course.


-Tobias

______________________________________________________________________

Tobias Kunze                     tkunze@ccrma.stanford.edu