13 Feb
2010
13 Feb
'10
11:16 p.m.
Just two general-purpose first order IIR sections
is all you need for either the forward or inverse
filter. Any textbook on digital filters will tell
you how to program them. Inverting the LF filter
requires an extra pole below the audio range to
avoid infinite DC gain.
I'm aware of how to construct digital IIR filters :-) I was hoping
you had a URL to a nice official analog topology. The specific
implementation details matter.
The channel EQ you'll find on most digital mixers
is
not linear-phase at all, nor acausal.
OK. Time to become incredibly overspecific:
Every digital EQ implementation I'm aware of for Linux is linear
phase. I wrote a few of 'em.
One could just build a digital equivalent of any of the old analog
topologies. For many filters (eg, compressors and the like) this is
totally the way to go. For EQ, I'll take a linear phase
implementation any day. That's the route taken by every piece of FOSS
EQ source code I've ever seen (it would not be surprising if I missed
a few). If you say VST has done a few this way (for whatever reason)
then I believe you.
In almost all
cases it's just first and second order IIR filters.
For plugins anything goes, but the most of them are
not linear phase, and no filter operating in real
time can ever be acausal - by definition.
Negative delays are perfectly possible in digital. Well, if you
ignore the wallclock (assume a global system latency, and a local
negative latency within the system). It's just a semantic/terminology
argument at this point.
Cheers,
Monty