Capturing Impulses...software, techniques.
-
@lalalandsynth I have to deconvolve multiple time to get a usable result (without changing any settings).
Sometimes I'll get an unexpected frequency response or ghost echos. -
@dustbro just did an IR long time ago and was good. Couple of tricks to get it working since it's not aimed to final user I guess.
-
So how does convolution work with analog gear? I assume you take impulses of every setting (or as many as you can) and then interpolate between them somehow?
That last bit always confused me
-
@iamlamprey there is no interpolation. Well there's a couple of model machines out there that take "frames" and apply them in different ways, but convolution is straight(but CPU intensive) maths on a fixed set of numbers (the impulse). So in most situations where the result is dependent on a changing input ( say a desk or a compressor or a tape machine), then convolution is probably not going to help much without first streaming your input(on say frequency or volume) and passing it into multiple convolutions - so expensive and all you did was move the granularity of the interpolation required ( and you still cant have interpolation).
-
@Lindon I see, thanks Lindon. I suppose it's better to use it as a base then recreate the "sound" with algos/chains instead? I'd still prefer that route to the whole "LTspice recreating the circuit" thing
sorry for pseudo-derailing
-
I am making some sweep impulses of reverbs and then use the Tick method to capture the "Colour" of the unit in question, I have yet to test if that method of capturing the colour is worth doing , I need to do some more experiments.
-
@lalalandsynth my experience is that the sweep will capture the colour anyway - your experience may end up different.
HISE Development for hire.
www.channelrobot.com -
@Lindon yes, today i am doing both as a precaution.
-
@Lindon actually I meant sweep for time based stuff..reverbs..and tick for comp/eq
-
Next up, amps
-
Then i need to figure out how to impulse mics
-
@lalalandsynth just stop now OK, I already decided I had gone off you because of all the stuff in the first set of pictures - the Crane Song stuff especially... but now the amps and mics? In a minute you'll post a picture of an early Rick 360 and I will never talk to you again...
Really my envy knows no bounds...
-
Lol, i will see if i can find one , there is a whole wall of guitars and bass here ;)
-
@lalalandsynth is this YOUR studio?
HISE Development for hire.
www.channelrobot.com -
@Lindon No, not my studio ,but I own a lot of the synths there.
I have a synth repair and sales shop in a room in the studio. -
@lalalandsynth ok well I'm very slightly less envious then...
-
Must collect them all!
-
So am capturing the EMT 244 today . it does have quite a bit of low end hums and noise , any thoughts on how to deal with that kind of thing , as it is the noise would be added to the impulse.
-
@lalalandsynth That's a super-nice studio! Regarding the noise from the EMT - you might want to capture a bit of that after the sweep and noise reduce the impulse in RX.
And when it comes to mics, this might be of interest to you: http://micirp.blogspot.com/p/about-micirp.html :-)
-
Looks like i am getting some aliasing ?
I should have tested a longer impulse to see if there was less aliasing , just to be sure , dont know if that would change anything ? Having said that , if aliasing is part of the sound then so be it :)This impulse is from the EMT 246 which really only goes to 12k (probably contributes to how nice it sounds) and the "aliasing" is pretty clear.
Also wondering if I should normalize impulses before deconvolving or if I should normalize the resulting impulse , guess I need to do some tests.
