Voxengo Premium Membership - All Voxengo Plugins For a Fixed One-Time Fee
Forums     Plugins     Impulse Modeler and Deconvolver Convolving gear


Hi ya,

I played a 4 second 20~15K sine sweep through a stomp box and recorded the result.

I then put the test file and the recording through deconvolver.

What I ended up with was a sort of synth envelope follower, whereas I was expecting more of an EQ fingerprint.

Did I do something wrong?

Should this process not work?

It's still a frequency plot through the time domain yes ?

The sweep seemed far more applicable than an impulse, especially as I was interested in the frequency response of the system under test.

But in theory you could convolve anything - including a delay line yes ?

Slightly baffled ;-)


BTW, as insane as it sounds, does anyone have a wave file of an impulse trigger- like a balloon or starter pistol they want to share ?

Presumably that would have to be a pretty anechoic recording ?

Or if you were going to synthesise it, could it be a few mS white noise burts or something ?

Should it have an envelope, or should it have attack and release and if so of what magnitude ?

Thanks.


I'm no expert, but I was under the impression that one could create an "ideal" impulse as single sample valued 1.0 ( 0 dBFS) followed by any number of samples value 0 (silence).  I.e. a one sample spike followed by silence.  Can anyone confirm this ?

midiwhale: Hi ya,

midiwhale: I played a 4 second 20~15K sine sweep through a stomp box and recorded the result.

midiwhale: I then put the test file and the recording through deconvolver.

midiwhale: What I ended up with was a sort of synth envelope follower, whereas I was expecting more of an EQ fingerprint.

midiwhale: Did I do something wrong?

Without hearing your results and having original files I can't tell you what's wrong.  Please send me test tone and recorded file to info@voxengo.com .


midiwhale: BTW, as insane as it sounds, does anyone have a wave file of an impulse trigger- like a balloon or starter pistol they want to share ?

I do not think this is the best option to capture acoustic spaces, because you would not have an option to deconvolve the recording afterwards while without deconvolution your impulse response will have a 'balloon pop' or 'starter pistol' sound imprinted into it.


ToneCarver: I'm no expert, but I was under the impression that one could create an ''ideal'' impulse as single sample valued 1.0 ( 0 dBFS) followed by any number of samples value 0 (silence).  I.e. a one sample spike followed by silence.  Can anyone confirm this ?

This can be used with hardware devices only.  Speakers just do not produce enough power out of a single spike to be captured precisely by a microphone.


Am I wrong in thinking that what the deconvolver does (put basically) is remove the test tone from the captured impulse ?

i.e. it removes the actual tone sweep, just leaving the reverb image ?

If this is true, if I give it the starter pistol as the relevent wave, would it not handle that equally as well as the sweep ?

So should what I did, have captured a normal reverb/space and/or a tonal device (ie EQ or retro gear)?

Or does retro gear need to be handled differently for some reason?

Aleksy, I'll email you the source files.


midiwhale: Am I wrong in thinking that what the deconvolver does (put basically) is remove the test tone from the captured impulse ?

Yes, deconvolution can be seen as 'removal' of test tone from the recorded file.  Test tone can be also seen as a 'carrier frequency' that carries impulse response (reverb or hardware device impulse response).

midiwhale: i.e. it removes the actual tone sweep, just leaving the reverb image ?

Yes, it does.

midiwhale: If this is true, if I give it the starter pistol as the relevent wave, would it not handle that equally as well as the sweep ?

It will handle starter pistol sound as test tone as well.  However, there is always a problem of having a 'dry' anechoic starter pistol sound.  Usually, it is impossible to get it.  I think that 'starter pistol' is a technology of the past.  It's not the best thing to cope with explosives and always fearing to run out of the ammo supply.

midiwhale: So should what I did, have captured a normal reverb/space and/or a tonal device (ie EQ or retro gear)?

If you are speaking about the file you've sent me, I do not even know what you should get.  That is too non-linear device: it does some volume ramping, repetitions and something like chorusing.  Such processes cannot be captured with linear convolution.

midiwhale: Or does retro gear need to be handled differently for some reason?

Not necessarily.  However, non-linear processes should simply not be captured as results won't be predictable.

This topic was last updated 180 days ago, and thus it was archived.  Replying is disabled for this topic.