The No.1 Website for Pro Audio
Removing small room sound via reverse convolution process?
Old 2nd September 2018
  #1
Lives for gear
 
jnorman's Avatar
Removing small room sound via reverse convolution process?

So we know that by creating impulse response wave files in a given space, we can use convolution algorithms to generate a reverb response generally matching that space when applied to a dry recording. I was thinking that perhaps it might be possible to create an impulse response in my smallish studio (17x23x8ft), and then, and using a similar process like when we apply subtractive EQ to remove hvac noise, it might be possible to use a convolution type reverb algorithm in reverse to remove the ambient signature of the room. Is the idea crazy? Is there, perchance something like this already out there? I know there is reverb removal software like de-verb, so why couldn’t we have a similar piece of software that uses an IR of the exact ambience that you wish to remove? Thanks for any thoughts.
Old 2nd September 2018
  #2
Lives for gear
 
bgood's Avatar
Quote:
Originally Posted by jnorman View Post
So we know that by creating impulse response wave files in a given space, we can use convolution algorithms to generate a reverb response generally matching that space when applied to a dry recording. I was thinking that perhaps it might be possible to create an impulse response in my smallish studio (17x23x8ft), and then, and using a similar process like when we apply subtractive EQ to remove hvac noise, it might be possible to use a convolution type reverb algorithm in reverse to remove the ambient signature of the room. Is the idea crazy? Is there, perchance something like this already out there? I know there is reverb removal software like de-verb, so why couldn’t we have a similar piece of software that uses an IR of the exact ambience that you wish to remove? Thanks for any thoughts.
That is interesting...
Old 2nd September 2018
  #3
Lives for gear
It might help, when applying it, to render it in real time (1x), in reverse ...... because then the 'vaccuum cleaner' you're applying hits the 'trailing edge' (ie the reverb) first....and it should be easier to tweak when you want the processing to taper-off, as it hits the 'meat of the sound' (sorry about all the 'syllogisms' implied....discard, if the idea itself is crap at base ?)
Old 2nd September 2018
  #4
Lives for gear
 
Bruce Watson's Avatar
 

Quote:
Originally Posted by jnorman View Post
I was thinking that perhaps it might be possible to create an impulse response in my smallish studio (17x23x8ft), and then, and using a similar process like when we apply subtractive EQ to remove hvac noise, it might be possible to use a convolution type reverb algorithm in reverse to remove the ambient signature of the room. Is the idea crazy?
I think you're making an interesting assumption that's not true: that the process of applying reverb to a dry source is symmetric to the processes of a de-reverb algorithm. And it's not at all symmetric. You can't just apply a "reverb algorithm in reverse to remove the ambient signature of the room" any more than you can apply a recipe in reverse and remove the salt from a cake.

What's wrong with getting a dry recording the old fashioned way -- treating the studio before you record?

But if you really don't want to treat the studio for some reason, what's wrong with buying a license for iZotope RX 6, or even just their de-reverb plugin? It's considerably more complex than "a convolution type reverb algorithm in reverse", and (so far) it's not quite as good as recording in a treated studio, but it's pretty darn good at lessening the effects of "small room sound". It might be sufficient for your needs.
Old 2nd September 2018
  #5
Lives for gear
 
Plush's Avatar
Once the small room sound is encoded, it can never be removed.

It is the death knell for good quality sound.

There is no way around it.
Old 2nd September 2018
  #6
Lives for gear
 

Desperate times call for desperate measures, and if you are in a client "situation" around this anything goes that you have time for. However, I would not recommend planning a recording around this technique from the get go--so much easier to get the right sound the old school way....
Old 2nd September 2018
  #7
Lives for gear
 
Plush's Avatar
No, it is a failure and calls for a re-record. Bill the client again.
Old 2nd September 2018
  #8
Lives for gear
 
robert82's Avatar
Quote:
Originally Posted by Plush View Post
Once the small room sound is encoded, it can never be removed.

It is the death knell for good quality sound.

There is no way around it.
This should be a sticky at the top of both the Acoustics forum, and the Newbie's forum. Minding this simple bit of wisdom would prevent a lot of needless frustration.
Old 2nd September 2018
  #9
Lives for gear
 

on individual tracks of individual close mics, applying (multiband) expanders can partially lower room impression.

but there is no such thing as 'reverse convolution'...

(next time, use either more directional mics, move them closer, move into a larger room or do all of this together)
Old 2nd September 2018
  #10
Lives for gear
 
jnorman's Avatar
So, you’re saying that software like de-reverb will not remove a small room signature?

My wife runs a flute studio in our studio and I am very regularly required to record large numbers of students for auditions, college applications, etc. the students can not afford renting halls or churches for this purpose, and I can’t really absorb the cost to myself to do so many remote recordings when I am doing my best to provide most of the recordings for free to financially disadvantaged students. I must figure out the most effective way to get decent recordings in the space I have. I do have acoustic treatment on the ceiling and plenty of diffusion on the walls. Thanks again for all your comments.
Old 2nd September 2018
  #11
Lives for gear
 
jnorman's Avatar
Hmmm - this is from the Acon DeVerberate page:
“The Early Reflections Filter uses sophisticated statistical methods to estimate the so called impulse response of the early reflections. The impulse response is then utilized to calculate an inverse filter that cancels out the effect of the early reflections.”

That sounds rather like what I was try to describe. Maybe I will try out the demo and see what it does.
Old 2nd September 2018
  #12
Lives for gear
 

Quote:
Originally Posted by jnorman View Post
So, you’re saying that software like de-reverb will not remove a small room signature?

My wife runs a flute studio in our studio and I am very regularly required to record large numbers of students for auditions, college applications, etc. the students can not afford renting halls or churches for this purpose, and I can’t really absorb the cost to myself to do so many remote recordings when I am doing my best to provide most of the recordings for free to financially disadvantaged students. I must figure out the most effective way to get decent recordings in the space I have. I do have acoustic treatment on the ceiling and plenty of diffusion on the walls. Thanks again for all your comments.

nice you're supporting financially disadvantaged students!

i'd use a single directional mic close to the instrument (plus additional expander if still needed) and then either two spaced mics at the far end of the studio (and/)or some efx device to add ambience/reverb.

i cannot comment on the 'de-reverb' plug-in; my comment was more on the possibilities and limitations of convolution technique - of course you can try to eliminate sound by applying various degrees of phase cancellation; tricky though...

(i'm sometimes using a speaker processor to do the opposite: aligning signals of distant mics by applying delay and phase inversions in multiple frequency areas).
Old 2nd September 2018
  #13
Lives for gear
 
jnorman's Avatar
Just tried demo of Acon DeVerberate - almost any application of the effect results in unacceptable artifacts. Not a solution for me...
Old 2nd September 2018
  #14
Lives for gear
 
robert82's Avatar
To the OP:

You cannot remove the comb filtering, standing waves, and frequency anomalies of your room. So eliminate them:

deaden your room even more (your space is too small to benefit from diffusion). Place microphones very close to the source. Use artificial ambience to offset the deadened acoustics.
Old 2nd September 2018
  #15
Lives for gear
 
Bruce Watson's Avatar
 

Quote:
Originally Posted by jnorman View Post
So, you’re saying that software like de-reverb will not remove a small room signature?
Nothing we have yet will completely remove it. But most of these apps can make a dent in it. The question is, is it enough? And for at least one of these you're saying "no". I hear/read that iZotope RX 6 is the best of the current lot, but my experience level with these things is low.

Quote:
Originally Posted by jnorman View Post
I must figure out the most effective way to get decent recordings in the space I have. I do have acoustic treatment on the ceiling and plenty of diffusion on the walls.
I agree with Robert82 about erring on the side of absorption. Better to make a really dead space and apply the reverb you want to the resulting recording, than to try to remove the reverb you don't want.

The thing of course is to use absorption that works best on the frequencies that give you the most trouble. Which is probably decently high in the frequency spectrum for flute players. But you know this already...
Old 2nd September 2018
  #16
Lives for gear
 

I once had a university DSP team give me a 'demo' of their "de-reverberation" software (aimed at making surveillance tapes more distinct). It did achieve significant improvement. Only much later did I find out that the demo consisted of an original recording that had then been artificially reverberated to make the 'subject' tape, and the two were played in reverse order. Fake news.

I have heard a box used by police forensics that could take two mics - one nearer the subject - and use correlation techniques to subtract interfering sounds. The box I suspect had its origins not million miles from Langley VA, and is unlikely to be available to the average sound recordist ...

One big problem is that you will construct a model for one room one microhpne and one instrument at specified positions. Change position even slightly and you start all over again ... A war-surplus Cray might be useful as the processing vehicle.
Old 3rd September 2018
  #17
Lives for gear
Quote:
Originally Posted by jnorman View Post
So, you’re saying that software like de-reverb will not remove a small room signature?

My wife runs a flute studio in our studio and I am very regularly required to record large numbers of students for auditions, college applications, etc. the students can not afford renting halls or churches for this purpose, and I can’t really absorb the cost to myself to do so many remote recordings when I am doing my best to provide most of the recordings for free to financially disadvantaged students. I must figure out the most effective way to get decent recordings in the space I have. I do have acoustic treatment on the ceiling and plenty of diffusion on the walls. Thanks again for all your comments.
Unfortunately the amount of deadening acoustic treatment required to 'neutralise the room acoustic' may distort the player's technique, since they won't be getting back the usual, expected reflections from walls etc that they are accustomed to ? The net result could be that they overblow to try and generate the same amount of 'room contribution' that they usually get at home or in the concert hall. So be attuned to any variations in their normal playing style that your treated room is prompting
Old 3rd September 2018
  #18
Lives for gear
 
emrr's Avatar
I feel like judicious use of the tools in RX6 are the best path once you've done what you can to the room. Things like subtracting the room tone and rumble for all frequencies below the fundamental of the instrument, etc. That stuff tends to present as part of the reverb, even though it isn't musically related, and isn't really excited by the instrument. Lots of little tricks that can add up to a big difference.
Old 3rd September 2018
  #19
Lives for gear
 

If there are a few problematic frequencies you
can use dynamic Eq to tame them.
Old 3rd September 2018
  #20
Lives for gear
Kill the room by really going to town with the dampening, then when you're recording them put cans on the player and feed them the most 'expensive' reverb sound you can find, via either plug ins or hardware reverb, from your mic feed.

Mono (or at least M-S, where you can reduce S for them) might be best...discuss and debate here...hearing themselves in stereo (and incl possible image wander) is an additional distraction they just don't need !

They'll sound like a million dollars, and their playing might improve accordingly too...the caveat being that they'll have to adjust to playing with cans on (probably one can off the ear (and muted in your send !) is best, to give some real world connection to their own playing.

Also, keep their own playback level lowish, so they don't hold back on dynamics.... because the mics/cans are giving them a louder (ie false) 'foldback' impression than they're actually playing

There will likely be a few players who can't handle the transition to cans, not sure what to do there....but what I'm suggesting is no worse than what session players worldwide have to do when overdubbing in a recording session. So it's just an aural adjustment issue really.

Open back vs closed back headphones...we can discuss that here also ?

Your wife can be the guinea pig for all this while you're experimenting, and if she hates it, you'll simply have to take the heat

The adjustment process is actually no worse than the discomfort of hearing your voice amplified through a PA system for the first time...how does that Robbie Robertson song about the Crazy River go:
She said "There's one thing you've got to learn
Is not to be afraid of it."
I said "No, I like it, I like it, it's good."
She said "You like it now
But you'll learn to love it later."

Hopefully you've got a few 'million dollar rooms' in your reverbs collection (be sure they're higher in the monitoring (cans) chain than they are in the recording...in fact, best not to print them at all to the recording, so you can trickle them in at the right level during edit time !)

Last edited by studer58; 3rd September 2018 at 04:06 AM..
Old 3rd September 2018
  #21
Lives for gear
 

Quote:
Originally Posted by jnorman View Post
I do have acoustic treatment on the ceiling and plenty of diffusion on the walls.
Try using 6" thick bonded acoustical cotton to completely cover all the walls and ceiling (and remove any rug / carpet
from the floor) combined with high quality reverb processing. (Correction: 2" thick is sufficient).

Last edited by aracu; 3rd September 2018 at 01:42 PM..
Old 3rd September 2018
  #22
Lives for gear
 

Quote:
Originally Posted by aracu View Post
Try using 6" thick bonded acoustical cotton to completely cover all the walls and ceiling (and remove any rug / carpet
from the floor) combined with high quality reverb processing.
This will tame mid and high frequeny issues but
but what about low frequency room mode problems?
Old 3rd September 2018
  #23
Lives for gear
Quote:
Originally Posted by Folkie View Post
This will tame mid and high frequeny issues but
but what about low frequency room mode problems?
A solo flute's gonna excite low frequency boundary modes...tell me more ? If you have no room mode problems below 247Hz you'll be fine. Here's a great guide to improving rooms: Room For Improvement |

Last edited by studer58; 3rd September 2018 at 09:24 AM..
Old 3rd September 2018
  #24
Lives for gear
 

You are right but if he’s going to the expense of proper room treatment, it’s better not to overdo
absorption of mid and high frequencies. In small rooms, low frequency room modes are important
to deal with.
Old 3rd September 2018
  #25
Lives for gear
Quote:
Originally Posted by Folkie View Post
You are right but if he’s going to the expense of proper room treatment, it’s better not to overdo
absorption of mid and high frequencies. In small rooms, low frequency room modes are important
to deal with.
Yes, it's better to aim for even diffusion and absorption across the audible spectrum, and that will likely entail some bass trapping too...a studio room acoustically tailored to one instrument alone isn't smart...what if a piano, cello or double bass drops in for a concert ?
Old 3rd September 2018
  #26
Lives for gear
Quote:
Originally Posted by aracu View Post
Try using 6" thick bonded acoustical cotton to completely cover all the walls and ceiling (and remove any rug / carpet
from the floor) combined with high quality reverb processing.
in the article I linked to in post #23 : "It is vital to understand that the low-frequency performance of foam panels and similar absorbers is proportional to their thickness: the thicker the absorber, the more effective it is at lower frequencies. A typical two-inch foam panel, for example, ceases to have much effect below 300Hz or so when fixed directly to a wall. However, its general performance, and low-frequency performance in particular, is improved if it can be spaced away from the wall by a few inches. In practice, a piece of two-inch foam mounted two inches from a wall is almost as effective as a piece of four-inch foam glued directly on to the wall.

The required thickness of foam or other absorptive panels (and the distance they are placed from the wall) relates directly to the range of frequencies that can be absorbed — due to the simple fact that the surface of the foam needs to be spaced around a quarter wavelength or more of the frequency being reflected from the wall to have the optimum effect. Lower-pitched sounds have wavelengths measured in tens of feet, so there's no wonder that a couple of inches of foam won't have much impact here!

Absorbers work efficiently where the air movement caused by sound is at its greatest velocity. For our purposes, that is at a quarter of the wavelength. Directly adjacent to the wall surface, there is no air movement, only pressure variations — so imagine, then, how futile it is to stick carpet all over a studio's walls and ceilings! The room might not ring when you clap your hands but only the highest audio frequencies are being absorbed.

This leaves resonances in the mid-range and bass end to predominate, and the result is a room that sounds boxy, honky and dull. Even when using properly specified foam or Rockwool absorbers, it is unwise to cover too much of the room's total wall and ceiling area, as you'll simply mop up all the mids and highs, leaving bass-end resonance problems. What you need is just enough treatment to stop the room sounding too live, combined with a practical range of treatments to ensue that you tame the low end to balance it with the damped mids and highs"
Old 3rd September 2018
  #27
Lives for gear
 

Quote:
Originally Posted by Folkie View Post
This will tame mid and high frequeny issues but
but what about low frequency room mode problems?
Just 2" thick bonded acoustical cotton, if completely covering all the walls and ceiling, makes it sound as if there are no standing waves (and no reverb), like recording in a quiet forest. The drawback is that reverb has to be artificial. It cannot replace a good hall for recording classical music.

I haven't tried recording an instrument with very low bass frequencies with this room treatment.
Old 3rd September 2018
  #28
Lives for gear
 

Quote:
Originally Posted by Bruce Watson View Post
I think you're making an interesting assumption that's not true: that the process of applying reverb to a dry source is symmetric to the processes of a de-reverb algorithm. And it's not at all symmetric. You can't just apply a "reverb algorithm in reverse to remove the ambient signature of the room" any more than you can apply a recipe in reverse and remove the salt from a cake.
Yes you can and it is called a deconvolution process. It's the exact inverse of a convolution process (what is currently used in commercial IR-based reverb plugins).

Theoretically, it should be possible to first record an impulse thru the mic doing the recording, then record, then remove the ambient reverb using the recorded impulse. It just happens that there's no available commercial product yet. I've not looked closely at the matter but I don't see any reason why it wouldn't be doable.

Of course, it wouldn't be a perfect process so reverb wouldn't be 100% removed, just like convolution reverb is slightly different from a real recorded reverb. I think it could produce interesting results if done properly.
Old 3rd September 2018
  #29
Lives for gear
 
mutetourettes's Avatar
 

I think it might be worth trying some zynaptiq stuff - is it unveil that does the reverb removal? unfilter? have a dig around their website/videos
Old 3rd September 2018
  #30
Lives for snowflakes
 
12ax7's Avatar
 

The only way I could see this idea working is if you made your own IR using the exact same "problem room" you're trying to "eliminate".

I would also think that the IR would have to be made using the same mic as the one used on the "problem track"; and that the mic would have to be positioned EXACTLY the same!

I think the wild card would be all the details of how you controlled the sound used to create the IR. (That would be the trickiest part.)

...Even then, I'm sure there'd still be artifacts.

(But maybe such a process might still be useful to some degree if done properly).

That's the only way I could see for it to really work.
.
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump