Go to bottom

Revision 2016 hall impulse response

category: parties [glöplog]
Hey there!

Ever wanted to simulate what your entry will sound like at the partyplace? Here is something for you.

Acryd and netpoet managed to create recordings of Dirac impulses, played through the PA, in the empty Revision hall. While the circumstances were not good enough for a "perfect" result, due to high noise floor and sub-par recording equipment, it still came out good enough to be usable. I just cleaned the recordings up and made a short demo of the jingle played through the impulse response (first clean, then with simulated hall).

Get the IR and the demo as long as it's available: http://www.jco.de/revision2016/revision2016_ir.zip

I case you have no idea what I'm talking about: https://en.wikipedia.org/wiki/Impulse_response

Sidenote: It's much too boomy, I applied a low shelf filter reducing everything below 100Hz about 12db.

If you wanna use it to simulate what your entry would sound like in the hall, use any random convolution reverb plugin of your choice and import the response. There are freeware plugins available as well. For the demo I used REVerence, which comes with Cubase. Again, this is not a perfectly accurate measurement, and it was created in an empty hall, but imho good enough for a solid impression. Have fun with it!
added on the 2016-04-07 16:03:05 by jco jco
Awesome :D
added on the 2016-04-07 16:16:31 by Preacher Preacher
Might've saved me some embarassment this year. Thanks! :)
added on the 2016-04-07 16:32:37 by Subi Subi
Fun stuff. I assume this is made from a frequency sweep? (You don't want to _actually_ record a Dirac pulse since you're then limited of the SNR of your system basically at that single sample.)
added on the 2016-04-07 16:33:50 by Sesse Sesse
Hint for the Native Instruments users: Komplete includes an IR plugin that is called "Reflektor" and that is, for some reason, in the Guitar Rig 5 package. I tried jco's file; it works.
added on the 2016-04-07 16:35:32 by netpoet netpoet
Fucking awesome :D, gotta try later.

(Now someone make one for Messukeskus / Hartwall too.)
added on the 2016-04-07 16:39:53 by noby noby
The impulse was an actual Dirac impulse which I also found confusing; would have opted for a frequency sweep, too. Had to turn up everything to the max to get it even audible enough.
added on the 2016-04-07 16:56:18 by kb_ kb_
Quite boomy indeed, but certainly useful! Tried it with SIR1, which is a nice freeware convolution reverb. Just remember to completely bypass the dry signal for a somewhat realistic result. :)
REAPER users can just use ReaVerb btw :)
added on the 2016-04-07 16:57:26 by kb_ kb_
kb: recording an impulse allows to capture all kinds of echo/answer effects along the time domain, not just the pure frequency response. I find this more interesting when huge halls are involved. If it's possible to reconstruct those kinds of effects from a sweep as well, plz let me know how.
for a good measurement of the actual frequency response, a proper sweep would have been more appropriate. also, it would have annoyed the poor orgas trying to finally get some sleep even more, I guess ;)
added on the 2016-04-07 17:12:33 by jco jco
jco, you can actually deconvolve the frequency sweep to get to a proper IR. Only real downside is that it allows non linear resonances (such as wiggly tables :) to get into swing more.


But eg. our audio tech actually measured phase response of the PA with a frequency sweep to set up the delays in the amps.
added on the 2016-04-07 17:30:21 by kb_ kb_
Thanks a lot!
added on the 2016-04-07 17:47:30 by cce cce
hah, great. Seconding Noby too. That would be awesome, maybe ask Spiikki if he can do it
added on the 2016-04-07 18:04:31 by leGend leGend
didnt el-bee do something like this for assembly once?
added on the 2016-04-07 18:08:03 by wysiwtf wysiwtf
Super stuff!
added on the 2016-04-07 18:10:44 by trc_wm trc_wm
ah, finally that almost-revision bass rumble out of my subwoofer :PPPP
added on the 2016-04-07 18:35:50 by Maali Maali
jco, you can actually deconvolve the frequency sweep to get to a proper IR. Only real downside is that it allows non linear resonances (such as wiggly tables :) to get into swing more.

The non-linear components can be separated from the linear response as Volterra series during deconvolution though. Sine sweeps can even be used to measure and simulate non-linear amplifiers.
added on the 2016-04-07 18:43:28 by absence absence
BB Image - almost like the real experience!! :P
added on the 2016-04-07 18:51:02 by Maali Maali
FWIW, the deconvolution for a linear sweep is utterly trivial—just convolve with the time-reversed sweep. (For a logarithmic sweep, you need a slightly modified variant.)

I'd guess not getting the nonlinearities is a small price to pay for actually getting the linear parts right. :-)

kb: Can your PA rig simply measure this itself and do the deconvolution? I have no idea whether that's common in PA these days, but it seems to work quite well for home receivers. (Of course, again, no luck for nonlinearities.)
added on the 2016-04-07 18:59:24 by Sesse Sesse
added on the 2016-04-07 19:08:48 by leGend leGend
Sesse: The reason it works well for home receivers is that they adjust it for exactly one specific listening position - and furthermore they don't deconvolve but just set up EQs and delays a bit. This is actually what we've done with the Revision PA this year, optimized for the area in front of the stage until about the middle of the hall. But you can't deconvolve a PA (...yet? :)), the impulse response looks vastly different for each possible position in the hall.
added on the 2016-04-07 20:09:55 by kb_ kb_
Oh my God!! That's so awesome!!!
added on the 2016-04-07 20:44:36 by djh0ffman djh0ffman
kb: My receiver allows you to set up a bunch of listening positions and then picks some sort of compromise between them. But I don't know if they actually mess with the phase, too, or just EQ + delay (+ volume for rear speakers) as you say.
added on the 2016-04-07 20:46:50 by Sesse Sesse
Lovely jco, thanks :D
added on the 2016-04-07 21:02:17 by ferris ferris
kb, sessse, absence: tnx will check all of this out.

have you configured those line arrays properly, btw? there was a very noticable drop in the heights starting somewhere in the mid of the hall...

related topic: i wonder how the sintifex stuff (focusrite liquid mix) managed to account for dynamic reponse. the kemper modeling amp is able to do that as well afaik. do they just blend between different impulses depending on current state of input? but then, how to decide which one to use when exactly? i could come up with some unscientific fakery, but kinda intrigued by what would be the accurate solution...
added on the 2016-04-07 21:37:11 by jco jco


Go to top