17 Apr
2003
17 Apr
'03
4:37 p.m.
On Thu, Apr 17, 2003 at 12:25:33PM -0400, Nick Tsocanos wrote:
I am going to make my Moog filter into a plugin as my
first attempt. I
think it sounds pretty good and the resonance is really good on it. I
modeled it directly from the analog-s-domain transfer function given by
stilson using a bilinear transform, and double sample it for stability.
It appears it only works properly if the coefficients are doubles.
Its notmal for filters to be unstable if you use floats internally, you
can sometimes get away with it if you clamp the cutoff.
I have figured out mostly what the problem was with
unitialized plugins.
The problem is I was not zeroing out my controlPorts when creating the
plugin, and it would come up with whacky values. Only some plugins don't
work correctly. Most work ok even with random values (but they make
really really strange sounds).
Yup, strange sounds are to be expected :) If they blow up then its a bug
though.
Also, I think that on some plugins, it expects input
to be normalized to
[-1,1] and, I am not clipping at the point I run the plugins, because I
am using a dither() algorithm which does the hard clipping when quantize
to 16-bit integers (for sound output).
Again, they shouldstill work, but might sound wierd.
Don't think I am criticizing any plugins!
No, please do there are a lot of undicovered bugs.
- Steve