Experimenting with FIRs

[steph_tsf]

This is an upgrade featuring the Analyzer.
The .zip file contains three files.
You need to load one such file upon running the .fsm. This defines the FIR filter. There is a "Load File" button for that.

[MyCo]

I've added the FIR coeff generator from Martin (from the Stream FFT project) to the FIR that I posted earlier... works quite well...

[steph_tsf]

I've added the FIR coeff generator from Martin (from the Stream FFT project) to the FIR that I posted earlier...

Wow! Your two FIR implementations (non optimized / optimized) look impressive. They both work on "packed audio" (four channels), applying a common FIR filter (defined by an external .txt file) to four audio channels. Actually, I need to apply a different FIR filter on each of two audio channels. Is there a speed penalty when using two "packed audio" FIRs instead of using two FIRs designed for single audio ?

[MyCo]

I wouldn't call it penalty, but if you don't use all 4 channels, you can optimize the code by removing those memory accesses for the unused channels. This is only possible in the assembler version.

[martinvicanek]

I've added the FIR coeff generator from Martin (from the Stream FFT project) to the FIR that I posted earlier... works quite well...

Cool.
This reference was mentioned in a recent discussion over at KVR regarding FIR Filter design. Below is an implementation in FS, it is quite easy to use and fun to play with. So far it only generates filter coefficients and shows transfer functions, I haven't plugged it into an actual FIR filter yet. Comments appreciated.

FIR_Design.png (97.43 KiB) Viewed 19415 times