Semi-Sweet Background Drone/Hum

[Jon Fairhurst]

I'm doing a corporate shoot about a laboratory project. The lab is terribly loud, so I recorded the researcher's description of the process first in a quiet area. On camera in the lab, I have him start a few sentences lip sync'ing a couple words music video style as he turns toward the workbench and I cut to other shots that don't show his mouth. To really sell that he is in the lab talking, I need to add a background drone.

So... what should I use for the drone? I can record a few minutes of "loud silence" in the lab, but it sounds fairly gross. But by keeping the levels low and EQ'ing it I'm thinking that I can get a pleasing background hum.

I don't want it to be as soft and pleasant as the sound of the bridge of the Enterprise, but I don't want it to be annoying and headache-inducing either.

So, will recording and EQ'ing the fans and such in the lab do the trick, or should I use a library sound or other technique?

Thanks!

[Battle Vaughan]

My take is biased by coming from a journalism environment, but our rules were always "it is what it is." No fakery, no fx, no substitutes, what we put out there must be the real and actual thing it purports to be. So my approach would be to record natural sound. If it is too loud or distracting, there's nothing wrong with lowering the volume, but the sound itself should be what it purports to be, I think. My two cents.

[Rick Reineke]

Using a noise reduction plug-in or application can likely attenuate the BG noise somewhat without losing the 'character' of the space. Using too much NR can and will cause annoying artifacts and do more harm than good.

[Jon Fairhurst]

Good point, Battle. I'll use the actual sound as a background and attenuate and EQ as needed.

It's really too loud for live recording. I tried to have him sync his initial words while I played his recording back on the speaker of the DR-100 at full bore. He could only hear his voice with the unit six inches or fewer from his head. He ended up rehearsing the cadence like that, we put the DR-100 away, and we went with it. With a bit of sound editing, I got his dialog to match the timing well.

Rick, NR isn't needed as I recorded the dialog in a quiet area. But to "put him in the space", I need to add noise.

Maybe I could get a Noise Addition plug-in. :)

[Battle Vaughan]

Because of the ethical concerns I would face in a news situation, I would start an interview with the man in an appropriate --- and quiet -- place and after establishing who's speaking, continue his voice as voice-over for the noisy bit, showing people working with some of the ambient nat sound under the voice track while that shot is running....this may not apply to you, I'm not being holier-than-thou, but I have only news experience and this is just my perspective....

[Jon Fairhurst]

This isn't a journalism project and the style doesn't try to make it look like a news piece, so I don't feel that the approach is a violation of trust. The lab location is where he actually does the work, so it's all good.

Unfortunately, putting the white noise of the lab under the voice sounds absolutely horrible. It just comes off as a bad recording with cheap preamps - even with extreme EQ. The only place it sounds good is when I bring up the noise as he opens the door to walk in.

I'll either need to find a sweeter pad, improve my processing of the voice, and/or put it over a light music bed. I don't need it to sound supernatural. I just need the audio to not distract the viewer from the story.

[Warren Kawamoto]

What do you see in the lab? Beakers full of bubbling liquid? If so, try an aquarium/ aquarium pump with bubbles. Bring the level way down in post. If no liquids visible in the lab, you could also try the hum of a window-mounted air conditioner, or refrigerator motor.

You may also find that you don't need any background efx sound if you add music under the narration.

[Vincent Oliver]

Set up an establishing shot with a wild track of the full background noise, as your narrator starts to talk then fade the noise down so it is still there as the backdrop for the production.

[Greg Miller]

Originally Posted by Jon Fairhurst

Unfortunately, putting the white noise of the lab under the voice sounds absolutely horrible.

I'm curious why you refer to it as "white noise." Does it really sound like a constant random white noise, that is non-changing in nature... very similar to the random noise produced by an electronic noise generator? (Hard for me to imagine a real-world lab sounding like that.) Or are you using the term "white noise" loosely?

It seems we're all making suggestions here, without having heard the actual noise.

[Jon Fairhurst]

Greg,

It's fan noise. The fans keep a positive pressure to keep dust out. There may be some exhaust fans in chemical hoods too, but I'm not sure. The fans don't have a whine, rattle, vibrato or other noticeable character. It just sounds like random noise to me. And, yeah, it's not technically "white" in that it probably doesn't have equal energy across the spectrum, but it's broadband noise. And when inserted in the video, it sounds just like a noisy preamp.

Warren and Vincent,

I think you're both on the right track. Fading the noise down after the initial moments works well. I have yet to try a music bed. I'll need to create some low key techno music and give it a try. I'm avoiding trying it with the wrong music so I don't get a false impression.

It's an interesting situation. It's where they really work. And it's loud in there. But their equipment is silent, apart from the sound of a laptop. And the fans that make the racket aren't in view. But anybody who's been in a controlled lab (we have to wear smocks, safety glasses, and booties - but not full bunny suits) knows that those locations aren't silent.

Part of the problem may be that I recorded the audio in a more damped, but not reflection-free room. I hung blankets as well. The room that they are in is all hard surfaces and no carpet. I should try re-recording some dialog in a more lively - but quiet - space to see if that fits better.

[Greg Miller]

Jon,

If I understand you correctly, you are mostly hearing the turbulence of the air moving through the diffusers, grilles, etc. If you're not hearing any motor noise from the fans, perhaps they are located at the other end of the ductwork.

So yes, I guess that might sound a bit like white noise, or perhaps filtered random noise that has some spectral distribution. And, if it's really like that, I suppose it might give the impression that it's a noisy preamp. In that case, I agree that Warren and Vincent both have good suggestions.

But what else is going on in the lab? Isn't there any equipment in use that creates an intermittent non-constant noise? Keyboards typing? Centrifuges spinning up and down (oh, wait, that was the Stuxnet track). Anything that would add a more "lab-like" sound to the BG? Footsteps? Doors closing? Cabinets opening/closing? If so, add it in.

Also, what's the distance from the camera to the talent in most of your shots? And how closely was the talent mic'ed when you recorded the tracks? Perhaps some of the incongruity comes from a big difference in perspective. If the talent is shown in MS most of the time, but was mic'ed at 6" in a dead room, that sound won't match the pix. You might be better if you re-record the voice track so it sounds a bit worse than it does now... less "studio-like" in quality. If the lab is noisy, and people normally talk loudly in there, does the talent talk relatively loudly on the track? Or was he soft and intimate because he was in a quiet studio setting? Deaden the studio a little, but not too much, back the mic off 2' or so, and have the talent project more when you re-record.

You're not striving for accuracy here... you're striving to create an illusion. Show biz... it's our life. ;-)

[Jon Fairhurst]

Excellent suggestions, Greg. Especially about the talent speaking loudly. I believe that I know of another room of similar size with similar hard floors that would be quieter. I can try recording in that location with a lav a bit low on the shirt and an omni a bit farther away than normal.

And you're right. It's about the illusion. I have dolly track and a small jib in the room and it's not like I show that and the crew. As a friend of mine used to say, "We don't want to present reality. We want to present what people perceive as reality." His example was of recording dialog in a small bathroom. It sounds horrible! But you don't want a dry sound either. You want enough, fast echo to sell the location, but it has to sound nice to the ear. That's how we imagine bathrooms to sound, not how they actually are.

I watched Ridley Scott's director's cut of Kingdom of Heaven on Blu-ray last night. (It's MUCH better than the original release, which was cut like an action movie. This one is long, but the stories play out naturally.) At one point an antagonist screams from a stone jail cell. The echo reverberates as if he's in a canyon! It's far from real, but it nails the feeling of the scene.

I don't need anything so dramatic. I just need the audience to watch, listen, and absorb the story without the audio feeling out of place.

[Richard Crowley]

It is common practice to record 30-60 seconds of "room tone" even in quiet spaces. That is what you use when editing to fill in gaps from editing, etc. Or for running it "under" the "voice-over" narration, etc.

I like the idea of a loud "establishing shot" showing the "context", not just the lab setting, but that is is very loud.

Have you tried a "headset mic" with an almost-invisible wand. It gets the microphone right at the talker's lips and maximizes the ratio between direct (speech) sound vs. ambient (noise).

I work at the facility where most of your CPU chips were designed. When we have to shoot in the clean-room fab, we have the same problem of a very loud environment (from the constant laminar-flow air-stream flowing around everything and everybody). But we have the advantage that humans must wear hoods or helmets that hide their mouths, so we don't have to worry about any "lip-sync" on camera, because nobody can see your lips in the fab. We could even dub it into bad Chinese and nobody would notice. :-)

[Jon Fairhurst]

I initially suggested the mask technique, and while they use masks for some parts of the process, they want to show the simplicity of the process, so masks are out. Oh well, I only show a few syllables before I cut away, so lip sync isn't a problem.

We have an E6 mic, but that won't necessarily solve the problem. I'd have the room noise baked in and wouldn't be able to process it separately. Again, we don't necessarily want to hear reality as much as we want to hear what we perceive reality to be.

Photographs are a great example. When standing in a room with fluorescent lights and a window, we don't perceive the room as being green and dark and the outside being blue and bright - but that's what an unprocessed photo will show. Through dual color balancing, shooting HDR and knocking down the outside scene, we can force the photo to look like we perceive the room when being there.

I listened closely to the bridge sound of a Star Trek Next Generation episode today. It's amazing how many modulated/phased sounds, pulses, and beeps there are within the drone. It's ever-changing and sounds alive. I think that, along with a believable early reflection on the voice are the keys to creating a believable, pleasing sound that is transparent to the viewer.

It's ironic that we need to process video and sound to make it feel transparently real. Excess reality can be distracting rather than invisible.