Filters on FV-1

[Vivek]

I'm just starting off on the FV-1 rabbit hole, please be kind.

I need to implement some audio filters in FV-1

Say a Cab Sim or a higher order LPF or a brick wall or SVF

I can imagine two routes

A. Biquad filter
I know the coefficients that I would need

Or

B. A fixed multiband parametric EQ where I change the FGQ of each band so that it adds up to the final response I need.


Which approach would be better ?

Or are there still better approaches ?


Where do I find sample code ? I know Spin site has multiband EQ while Digital Larry published some biquad filters. Any others ?


After I write the code, what's the best way to see a frequency response graph ?

[Rob Strand]

I don't use the FV-1 at all but if you know what you want the Biquad filters (which I'm reading as generalized second order IIR section) can pretty much do anything.   A very common audio processing scheme is to just cascade a stack of Biquad filters.

Equalizers are a bit hit and miss because you don't know what they are actually doing under the hood.   They are convenient to tweak when you don't know what you want.  However if you took the final equalizer control settings and transferred them to a set of Biquad filters the response could be way off due to details about how the equalizers are implemented.

When I experiment EQ'ing stuff I used equalizer blocks that *are* cascaded Biquads.  So when I take the final settings there's no doubt they will match the filter in your mind.   Well ... up to the point where analog filters and digital filters deviate as you approach the Nyquist frequency.      That's only an issue if you are thinking 100% in terms analog filter parameters.  Some EQ blocks try to remove the response distortion effect using a well known EQ design method which makes the digital EQ more like the analog EQ as you get near Nyquist.

No doubt some of the equalizer libraries will be using one of those two method but there are other schemes - you just don't know what to trust unless the documentation spells out the implementation.

[Digital Larry]

You can brute force implement biquads, though Spin ASM includes built in filter instructions that implement a 1 pole filter in 2 instructions, while the most common type of resonant filter you see throughout the examples is a state variable 2 pole structure that takes about 6 or 7 instructions and has control inputs for frequency and Q (if you need that).  Biquads by comparison are relatively a pain, however there may be some advantage to one approach or the other (e.g. filter accuracy, resistance to overload or stability at higher filter center frequencies near fs/2, which probably does not apply to a cab sim).