New way of cabinet impulse extraction

[deLuther]

Definitely not the best sound (as mentioned in video), maybe shitty, think of it just like about technology demonstration. This amp was chosen just for simplicity of demonstration.
Maybe I should play something with screamer to get tighter sound, but it`s not so principal.

What do you think?

 

[~BURNY~]

Didn't bother watching the whole stuff but here's what I do (which I think is quite similar but skip the power amp section coloration entirely):

1 Play any DI through an amp using your regular settings. A small 10s section is enough, preferably something busy with some palm muting to get a good representation of the low end.

2 Record the mic'ed signal on track A

+ Re'cord the speaker out signal (with an appropriate DI that can handle this type of signal, for example Type 85 inserted between the head and the cabinet) on track B

3 Export both tracks separately making sure they are the exact same length.

4 Open Voxengo deconvolver and use the speaker out signal as a test tone and the mic'ed signal as a target.

Check MP transform and silence (I think 00.4 sec is a good enough length for accuracy)

5 Deconvolve. Done.

I think Wild Hades of Ignite Amps came with the idea first.

 

[deLuther]

I think that Voxengo Deconvolver using Fourier transform for whole file, while MatcherCL uses averaging (like Ozone or CurveEQ, but with larger block size), sliding frame with KBD window (Welch analysis). There are no special requirements for files, just sample rates must be the same. I.e. MatcherCL is replacement for Deconvolver

 

[deLuther]

BTW, here is distribution of MatcherCL: http://forum.guitarplayer.ru/index.php?action=dlattach;topic=319541.0;attach=1073010
Use as you want, report about errors, suggestions are welcomed, enjoy.
For additional information about usage see readme inside archive.
Also software can be used as general spectrum matcher (no need to use Ozone, CurveEQ an perform deconvolution after).
Hoping that someone can post better example of usage

 

[~BURNY~]

Oh, didn't realize you were involved in the project, are you? Sorry that might have come off as rude.

 

[deLuther]

Oh, didn't realize you were involved in the project, are you? Sorry that might have come off as rude.

No problem

Video was made to explain how simple process of such deconvolution can be for end user, so more people can made impulses from interesting gear, while conventional approaches usually frighten most users.

 

[deLuther]

Another "little" demo of usage:

 

[FF666]

Humm... virus detected
http://www.avira.com/en/support-threats-description?tid=7815&tlang=en

 

[deLuther]

Maybe try this one link: https://dl.dropboxusercontent.com/u/3630120/MatcherCL.rar
But:
https://www.virustotal.com/ru/url/2...2109581d461233d0822520ff/analysis/1413524761/
and
https://www.virustotal.com/ru/file/...a225744936f4f8370e1a392b/analysis/1413524638/
(including Avira)
I`m personally using KIS and, of course, don`t have any problem. I`m don`t incorporated any kind of suspicious behavior, maybe only some bugs not detected yet

 

[FF666]

Haha I trust you, but it's what avira says.
I'll try it, looks an interesting tool.
Thank you!

 

[deLuther]

FF666
Do you updated Avira bases, as recommended in case of false detection?
You can send executable for detection, https://analysis.avira.com/en/submit to be sure.


Anyway, I`m updated distribution, some fixes were made, recommended for update.
http://forum.guitarplayer.ru/index.php?action=dlattach;topic=319541.0;attach=1073469
or
https://dl.dropboxusercontent.com/u/3630120/MatcherCL.rar

 

[deLuther]

New version, changed command line parameters handling, added some parameters to play with (see readme).
Link deleted due to bug in multichannel convolution.

 

[Pickasso]

Thanks for sharing your program, deLuther. I want to give it a try. In the video it sounds amazingly accurate to the source sound. How is it different from Voxengo Deconvolver? Are the files deconvolved in a different way?

 

[deLuther]

Thanks for sharing your program, deLuther. I want to give it a try. In the video it sounds amazingly accurate to the source sound. How is it different from Voxengo Deconvolver? Are the files deconvolved in a different way?

I`m explained difference before:

I think that Voxengo Deconvolver using Fourier transform for whole file, while MatcherCL uses averaging (like Ozone or CurveEQ, but with larger block size), sliding frame with KBD window (Welch analysis). There are no special requirements for files, just sample rates must be the same. I.e. MatcherCL is replacement for Deconvolver

MatcherCL works in the way like Ozone or CurveEQ matching, but has larger processing block (can be changed) and produces impulse directly from spectrums, in double precision. You can set version 0.2 to have blocks larger than input file, in that case it can be more similar to Deconvolver.

 

[deLuther]

https://dl.dropboxusercontent.com/u/3630120/MatcherCL_v0.21.rar
Multichannel (stereo and greater) convolution bugfix.

 

[deLuther]

https://dl.dropboxusercontent.com/u/3630120/MatcherCL_v0.22.rar
Fixed problem with very large analysis block sizes (if you use -bsa key), from 17, i.e. 2^18 and so on.
Not very critical, but it will be better to use this version.