Room Sampling For Convolution Reverb

Posted: Sat Apr 12, 2008 9:27 pm

can someone please go over the process of sampling a room to use in a reverb program like waves IR-1? Step by step, maybe?

Posted: Sun Apr 13, 2008 4:32 am

PC or mac? If you're on a PC grab the Voxengo Deconvolver app from here (you can use it for free if you're happy only to deconvolve 3 files per session, or buy it for $40). If you're on a mac... hopefully someone else can suggest something.

1. Use Deconvolver to generate some test tones. I suggest generating a few of different lengths, some with fades for the 'reversed' technique and some without. (read the help file)

2. Set up the best speaker you have available in the room you wish to sample.

3. Set up a stereo pair of mics at the other end of the room (or in the middle, or wherever you think would sound good for a conventional recording)

4. Play your test tones into the room while recording the mics, and don't press stop too quickly after each test tone has finished: you need to record some 'silence' after each. If you want you can do this several times with the mics (or the speaker) in different places.* If you use more than one test tone, make sure you name the files to indicate which was used.

5. Load your recordings into Deconvolver, along with the appropriate test tones. Set the required bit depth for your impulses, make sure you tick 'Reversed Technique' if the test tone used fades, and hit 'Process'.

6. Load the resulting IR wav files into an audio editor and carefully trim the silence from the start and end. You may need to experiment a bit with different lengths: too short and you risk damaging the reverb tail, too long and you will waste cpu resources, as the longer the IR file the more number crunching is required to convolve it. You may also wish to normalise the files if you didn't do so in Deconvolver.

7. Load the imulse(s) into Waves IR (or SIR, or Pristine Space or whatever it is you use) and enjoy!

* if you record one set of files with the speaker on the left side of the room, then move it over to the right side of the room and repeat, you could create sets of true stereo IRs. Similarly, moving the mics from the front of the room to the back might provide accurate surround reverbs.

Posted: Sun Apr 13, 2008 9:04 am

15 second sweeps are the optimal length. Beyond that, SNR benefits are a diminishing return, and slight air pressure variations can cause smearing.

Posted: Sun Apr 13, 2008 11:21 am

Aram over at Audio Ease - they make Altiverb - also recommends using a starter pistol if possible. Very loud sounds with sharp transients behave much differently than sine sweeps, etc.

Posted: Sun Apr 13, 2008 8:30 pm

cool.. thanks so much.. the deconvolver program makes sense once I look at it.. input test tone and recorded mic.. then it can use some form of common mode rejection to take out that test tone, maybe. thanks again!

Posted: Mon Apr 14, 2008 9:34 pm

ok.. one more thing.. in voxengo.. I generate a 15 second sweep and when I play it.. no matter from what part it starts with a "click" sound.. like a square wave type click.. should I worry about that? do others hear that? I could send or upload the sweep that I am getting if someone wants to see if they could figure out if I am doing something wrong..

Posted: Tue Apr 15, 2008 12:59 am

There is a feature that lets you generate a sweep with a fade in/fade out. Try that.

Posted: Tue Apr 15, 2008 1:47 am

You should only use the fade option when using the reversed deconvolution technique. I suggest trying both ways.

This topic sponsored by: