impulse response convolution

[aombk]

Hello,

I try to make an impulse response convolution effect for small Impulse response files (less than 1024 samples)
small files like this are used for simulating guitar cabinets and contain detailed EQ data (mostly) and a tiny reverb tail.

i modified an old synthmaker file called 3386_Convo 8000 - 07.osm

i also modified ConvoRev6c.fsm (by Martin Vicanek and the one modified by KG) that i think is better but it (not only the modified, the original too) produces a strange sound (like low bit resolution sound) every other time something else (like a switch for example) changes

but anyway. the thing is that the first schematic doesnt work properly. i compared their results with other commercial or free convolvers like ignite amps nadir or le pou lecab2 etc and the results are not the same. it seems to mess things up, especially high frequencies

the second schematic results seem closer to other convolvers, but it has that low bit resolution sound every other time.

can anyone help with this?

thanks.

[martinvicanek]

Try this schematic. It is derived from ConvoRev7.fsm with the IR size set to 1024.

[aombk]

ok i tried it.
check out this schematic.
i removed the dry - wet, delay and filter knobs.

just turn the switch on and off a couple of times.

[martinvicanek]

Okay, I see your point. In your last schematic, when switching on/off it seems to work correctly only every other time. Weird. The selectors in FS make the whole schematic recompile while it is still processing, which may lead to unexpected results. I replaced the selectors with soft variants and the problem is gone. The downside is that processing takes place even if you don't need it.

[aombk]

thanks, although i cannot make use of the soft selectors solution.

but why the convolver recompiles correctly every other time?
btw, the "Direct Convo 32" doesnt produce a problem with FS selectors.

so you dont think there could be another workaround for this or maybe some change in the FDL assembler code to avoid this problem?

[martinvicanek]

I think I finally fixed it. The issue was when you use the switch, FS will restart the audio and possibly recompile the entire schematic. It will not, however, go through the declarations, so when you have something like

Code: Select all

a = 0;

in the preamble, this will be ignored. Instead, FS will resume operation (starting at stage 0, mind you) with whatever value "a" was when you switched.

Have fun!

[aombk]

great, yeah, that makes sense. thank you very very much