Strange that you are saying this, because things might be a bit more complex than seemed to be just a few days ago ...partly because obtaining a perfect zero crossing is far from a simple matter.
Reading about filters didn't trigger anything until a few moments ago when I stumped on the IIR filters mention above, then the whole image came as a whole: "could this filters talk be related with that (early 2000s) discussion about IIR vs FIR crossovers/anti-aliasing filters quality?"
Being very much interested in all tech advances in pristine audio quality. Those early millennium times when Digital audio was severely being deprecated along the lines of early digital cameras versus film cameras, every good technical argument or advance in DSP was very much welcomed, specially if it was acce$sible.
The problem with the filtering was mentioned back then by several high profile DSP designers in SOSpubs and other interviews usually with obscure or passing references like inherent problems with identifying when a wave is above or below zero and some extra jargon I am far to decipher or fully comprehend, but if that could translate in better sound I was all in.
The wisdom then was that FIR filters were superior to IIR ones but needed more processing. So when I found the XILICA novel implementation of FIR filters, I was interested. Their XD series of speaker processors effectively implement these allegedly better filters in hardware product, for first time in the industry afaik.
However, and here is where the plot thickens, FIR filters offer the designer a choice of coefficients (Xilica's come fixed), and some solutions tried to offer customer choice on them. Here is an Audio forum thread (from 2010) that mentions cases:
https://soundforums.net/community/threa ... ents.6848/
Including the info of a dedicated software -the FIRmaker- already then able to help design your own FIR filters. http://firmaker.afmg.eu/
So latency is the name of the game; Firmaker page states "Generally FIRmaker Technology can be employed using any FIR-capable DSP hardware. This is independent of sample rate and FIR filter length. Tests have shown that a filter length of only 250 taps provides reasonably good results down to approx. 300 Hz. Depending on the actual setup, AFMG recommends a filter length of 500-1000 taps."
A dedicated forum gives formulas for different approaches to the question of how many taps are needed for what freqs: https://dsp.stackexchange.com/questions ... ilter-need
Anyway, not sure if something of the above is applicable in MFS. It is an interesting side of things didn't expect to come out...
