Jump to content

Room tone caliburation


curleysound

Recommended Posts

I work on a stage shooting hosts all day. We've had the stage for a long time, and it's been an acoustic Nightmare for a long time. Our set is a one to one recreation of John Lautner's "Chemosphere" house, which is up the hill across the 101 from the Universal tower. Willie Burton might remember it from Charlie's Angels, where Drew Barrymore got shot out of the window. The house essentially looks like a UFO crashed into the side of the hill, and our set is 1/2 of the disk. The open end faces our electric room, which is on the stage and has an open ceiling to allow the hot electric dimmer packs to vent. The issues are that the dimmer packs have loud fans, and the set is like an opera hall in reverse, concentrating outside noise inward to the set.

I temaed up with our post audio people to help the situation out today by rolling a room tone test. I took our trusty Sanken COS-11 (Lectro sm) out to the set, and rolled room tone in ten typically shot spots on the set, and then did the same with the night time mode on (which eliminates lots of dimmer hum). We're planning on running these through a spectrum analyzer and dropping out offensive frequencies. Hopefully this will allow us to have a quieter set until we move.

I'm also working on a plan to put sound dampening between the room and the set without affecting the venting, but this hasn't been approved yet. I love corporate mechanics.

Has anyone tried this type of error correction before? Have you had success with it? It's basically a band-aid on a problem that shouldn't exist, but I didn't build the place, I just work here :)

Cheers,

Tom

Link to comment
Share on other sites

There is a new roomtone technique. I confess to only peripheral familiarity with the concept but the idea is to do a computer mapping of the acoustic environment. After recording a reasonable amount of quiet audio one has someone snap a slate closed (one clap only) to make a reverberant noise. The computer is able to track the decay of the reverberation to model the room. Ideally, instead of clapping one would play back a frequency sweep through a speaker of known characteristic but this would seem to demand a level of cooperation rarely seen on a movie set. The slate clap is a short cut varient. Supposedly post, if they are equipped with the appropriate software and take the time, is able to more effectively clean up audio recorded in a room they have accoustically mapped. But I have no personal experience to vouch for any of this.

David Waelder

Link to comment
Share on other sites

There is a new roomtone technique. I confess to only peripheral familiarity with the concept but the idea is to do a computer mapping of the acoustic environment.

David Waelder

This is a technique, often referred to as "sparking a room" (because one method actually involved an electrical spark as the impulse) which produces a convolutional reverb, a sort of audio signature of the space. These models are also called "IR's" or Impulse Response models. In theory, there is then a computer "model" of the characteristics of the room in which you have recorded production sound. I have never known anyone to make use of convolutional reverb to aid in eliminating unwanted noises in a room. The most common use, and this goes for the hundreds of impulse models that are used in post sound, is to help in matching ADR to production. Rarely have the actual rooms or sets that were shot on been "sparked" but rather there are numerous model spaces that the editor can use to create the sound of the room. I don't pretend to know a great deal about this --- there is a good explanation in this MIX Magazine online article:  http://mixonline.com/recording/applications/audio_audio_ease_altiverb_2/  Altiverb (from Audio Ease) has done more work on this stuff than anyone else I know.

See also: http://en.wikipedia.org/wiki/Convolution_reverb

Again, I could imagine, as David does, that an IR made in a room with all the offending noises present, could be useful in fine-tuning this idea that you can separate out what you don't want and be left with what you do want. We all know the difficulty in obtaining even 30 seconds of useful room tone on the set, it is highly unlikely that we would be given the time to spark a room (which is actually a fairly involved process). I have never heard from any of my friends in post that they have employed IR's and convolutional reverb to deal with these specific problems of noises on the set.

Regards,  Jeff Wexler

Link to comment
Share on other sites

I once heard of a technique that involved the placement of two identical mics, one for the desired sound and one pointed at the offending noise in order to phase it out. Theoretically possible I suppose gioven the time to experiment with placement and level balance.

Link to comment
Share on other sites

Jeff, with regards to Impulse Responses, have you ever been approached by the Post Supervisor to gather I.R.s on set.  I know that gathering the impulse responses has evolved into a fairly elaborate ordeal; with sine wave sweeps from omni directional speakers and multiple locations, but the deconvolving software is also improving.  I wouldn't  be surprised that in the future, I.R. technology is used for all manner of acoustic/phase related problems.  By the way, lots of good info about Impulse Response technology at the Waves (PT plug-ins) website.

Travis Groves

Savannah, Ga

Link to comment
Share on other sites

I once heard of a technique that involved the placement of two identical mics, one for the desired sound and one pointed at the offending noise in order to phase it out. Theoretically possible I suppose gioven the time to experiment with placement and level balance.

This idea had some currency in the 1970s ( the famous Grateful Dead "Owsley" PA system used two mics per vocalist, wired out of phase, in an attempt to control leakage from the PA system, all of which was BEHIND the band and looking right at the mics).  It was suggested to me several times back then, so I tried it.  For normal location sound, drama or doc, it really does not work--there are too many variables and the resulting sound is very strange.

There have been many attempts to do audio NR based on "room fingerprinting", either by getting extra samples of roomtone or (more commonly) by pulling a piece of room from a quiet point in a scene.  Here's a secret I've learned from MANY years of doco verite sound noise reduction in post: you have to "fingerprint" and separately NR each shot.  Again--too many variables for a "one size fits all" approach if you really want to get the most out of the NR w/o getting artifacts.  This last, the artifacts introduced by all these systems, esp the software based ones, are the true limits of what can be done with a noisy scene.  Many times I will work and work and work a scene and get the noise down a good ways, and then compare it with the original and stay with something closer to that--living w/ the noise.  The artifacts and the software have a tendency to suck the life out of recordings, and I've become more of a minimalist about those tools as I've gotten more experienced.  The Big Daddy of location sound NR is Cedar, with a deserved reputation for doing useful and good sounding NR quickly (and even live, on some TV shows).  But it too can be overused.  BTW--if you want to know how a Cedar system will deal with a particular audio problem, you can send Cedar Inc a short sample and they will run it thru their systems and send you back the result, free.  (www.cedar-audio.com)

I agree that constant noises on the set are better than "event" noises, but getting rid of those constant sounds has a price too--usually in how lifelike the resulting audio is after the noise is removed.  So it is better to try to get those HMI ballasts etc out of the room if you can....

Philip Perkins

Link to comment
Share on other sites

Guest klingklang

This is a technique, often referred to as "sparking a room" (because one method actually involved an electrical spark as the impulse) which produces a convolutional reverb, a sort of audio signature of the space. These models are also called "IR's" or Impulse Response models. In theory, there is then a computer "model" of the characteristics of the room in which you have recorded production sound. I have never known anyone to make use of convolutional reverb to aid in eliminating unwanted noises in a room. The most common use, and this goes for the hundreds of impulse models that are used in post sound, is to help in matching ADR to production. Rarely have the actual rooms or sets that were shot on been "sparked" but rather there are numerous model spaces that the editor can use to create the sound of the room. I don't pretend to know a great deal about this --- there is a good explanation in this MIX Magazine online article:  http://mixonline.com/recording/applications/audio_audio_ease_altiverb_2/  Altiverb (from Audio Ease) has done more work on this stuff than anyone else I know.

See also: http://en.wikipedia.org/wiki/Convolution_reverb

Again, I could imagine, as David does, that an IR made in a room with all the offending noises present, could be useful in fine-tuning this idea that you can separate out what you don't want and be left with what you do want. We all know the difficulty in obtaining even 30 seconds of useful room tone on the set, it is highly unlikely that we would be given the time to spark a room (which is actually a fairly involved process). I have never heard from any of my friends in post that they have employed IR's and convolutional reverb to deal with these specific problems of noises on the set.

Regards,  Jeff Wexler

An IR is of no help when the purpose is to get rid of the noise on set. In fact a noisy enviroment will create a noisy IR unless you use MLS signals to make the IRs. I´ve done many IR recordings and use these kind of processors every day. One thing it clearly doesn´t do is to get rid of reverb or noise. An IR is a "fingerprint" of a system at a single "time-slice" it doesn´t cover dynamic changes in time.

To the original poster:

Finding the problematic frequencies by recording a room-tone might be a little help but this could be done with every single take you record during the shoot.

An EQ will be of limited help if the noise is a broadband-signal. You will have way better results using professional de-noisers like Sonic NoNoise, Waves X-Noise or even broadband expanders like the Dolby cat. 43 and 430 or the infamous Cedar DNS1000

If the noise-floor is very high these tools leave severe artifacts on you dialogue. Your first and best solution is to do everything possible to move those fans away and if that´s not possible to build something that will shield the noise.

After having done that it´s good to know that some tools MIGHT get rid of the noise. A lot can be done in post but just sending in a noise sample will not be of much help for post because you need the dialogue WITH the noise otherwise there is no way to judge if the noise will be a problem to remove without affecting the dialogue. Modern de-noisers just need a few milliseconds of the noise to de-noise. No need to record minutes of noise-samples. During the shoot you´ll be using different mics and the boom will constantly move so the sound of the noise on your recordings will constantly change. In post you use a tiny slice of noise from the original take. A tiny bit between the slate and "actian" can be enough if there´s not too much crew noise. So one static roomtone recorded days before the shoot will not really help.

What could heklp is to record some dialogue send that to post and the director and tell them that this is the way the entire scene is going to sound if they leave those dimmers on set. Maybe then they will move them away ;-)

Link to comment
Share on other sites

Guest klingklang

I once heard of a technique that involved the placement of two identical mics, one for the desired sound and one pointed at the offending noise in order to phase it out. Theoretically possible I suppose gioven the time to experiment with placement and level balance.

One of the most common "urban myths" of sound recording.

Link to comment
Share on other sites

curley,

my buddy just last week was shooting close to the genis and a row of computers that had to stay on for the shot, they both were set up down a narrow hallway that "shot" the humming and rumbling of the 2 sources straight to the set. so during the pre light he had production rent a "drummer plexiglass barricade" from sir in hollywood. and set it up in front of the hallway arch and then did the old trick with sound blankets on the c-stand arms. the plexiglass is only like 4' tall so he had to utilise the blankets on stands. (from the time of call to setup was 1hr) it cut off the "noise" that was killing his shot.

if you will be shooting at that ufo shaped set for the next couple weeks, you might want to set up the plexiglass divider.

good luck curley.

Link to comment
Share on other sites

Thanks all for your helpful responses. We're shooting about 50 wraparounds for the videos on our network every day. These wraparounds  are married to the videos, and put on a playback server. They are not run through post, and go directly to air, sometimes only an hour after being shot. They are laid over with bed level music live on air. Basically We're gearing up to go for a Furniture pad barrier between the open ceiling of the electric room and the set. It should bring the noise floor down a good 6 to 10 Db and we're going to try and notch out anything that remains. So we're looking at a (compressed) s/n of 42-46 Db versus 36 Db now. I'll post pics when I get the chance.

Thanks,

Tom

Link to comment
Share on other sites

Thanks all for your helpful responses. We're shooting about 50 wraparounds for the videos on our network every day. These wraparounds  are married to the videos, and put on a playback server. They are not run through post, and go directly to air, sometimes only an hour after being shot. They are laid over with bed level music live on air. Basically We're gearing up to go for a Furniture pad barrier between the open ceiling of the electric room and the set. It should bring the noise floor down a good 6 to 10 Db and we're going to try and notch out anything that remains. So we're looking at a (compressed) s/n of 42-46 Db versus 36 Db now. I'll post pics when I get the chance.

Thanks,

Tom

You seriously might want to experiment with a rented Cedar DNS1000 in this case, since there is no post production and the sound should be fairly consistent.  The direct methods (dealing with the disease and not the symptom) are best of course, but I've heard of live TV shows using the Cedar when in noisy situations as well.

Philip Perkins

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...