Silence, the great abyss. The infinite chasm of stillness that becomes deafening through patience… - An ancient Carbo proverb.
Let’s take a trip to a special place. A place where physics and consciousness intertwine to create a blissful state of aural euphoria. It’s a place we may never get to but our efforts toward this reticent destination paves the way for a strong foundation towards our common goal. That common goal is good sound on set.
As the acoustic ecologist Gordon Hempton once said, “Silence is not the absence of something but the presence of everything.” If you’ve ever tried to remain silent for 30 seconds for room tone you’ll likely relate to Hempton’s words. Silence can be incredibly hard for some individuals. Often, those that think they’re being silent are merely being quiet, or at least attempting to be. Yes, there’s a distinction between the two. The differences of which can have a major impact on how difficult of a job you’ll be giving to your dialog editor. What may seem like the faintest whisper to your colleague can end up ruining a great piece of dialog or alter what would have been an otherwise smooth edit. Sound is money people! It helps to imagine that our microphones are giant ears that listen, record, and save every single audible moment for posterity.
The human ear is a wonderful organ. It’s a complex machine of bones and hair cells that converts vibrations of air molecules into electrical signals that are sent to the brain, processed in the auditory cortex, and finally transformed from nerve impulses into intelligible symbols learned throughout life in order to navigate our way in the world. Our ears have an incredibly large dynamic range. The distance between the softest sound we can perceive and the loudest we can hear before damage is around 140db. Without getting into the math behind it, the decibel is a convenient way to abbreviate the measurement for changes in air pressure. The more scientific method for measuring sound would be to use Pascal, which involves a lot more numbers. For perspective, consider the average decibel level of common everyday sounds. (see chart below)
Image courtesy of the American Academy of Audiology
The threshold for hearing in a young, healthy human is 0db (or 0.00002 Pa if you prefer Pascal). Breathing at a normal rate registers about 10 decibels, which is 10x greater than what is colloquially known as an absence of sound. On set, with some notable exceptions, we spend most of our time dealing with the “moderate” and “loud” volume areas in the graph. This is where the dialog between actors will be. This decibel zone is where us production sound mixers are transforming your art into digital 1’s and 0’s. Following the decibel’s logarithmic scale we find that the volume of our much coveted dialog sits at a whopping 1 million times greater than 0db.
Ever wonder where sounds you previously thought were silent stack up on the decibel level?
A vibrating cellphone sitting on a table about 3 feet from away will register about 51db. Shifting your weight while seated in a fabric chair (no metal or wood creaks) hits about 37db. Walking on a creaky wooden floor can be anywhere from 48db to 60db or higher depending on how a person walks.
This is all relative of course. The average “room tone” of an ideal location for a production shoot, practically speaking, is about 20db-30db in volume. That’s a room without HVAC bellowing from the ceiling of a corporate office building and no traffic outside seeping in though the windows. Ever been in a seemingly quiet room and your production sound mixer complains about a noisy hard drive fan that needs to be turned off or moved? Well, that’s because that hard drive fan can be as loud as 35-45db just in it’s resting state. The goal when choosing a shooting location is to find a space where the ambient sound level is drastically lower than the average level of a person speaking in a normal context. This creates what is referred to as a good SNR (signal-to-noise-ratio). When we have a good SNR we are able to edit the recorded spoken word with greater ease. There still may be some lingering broadband residual noise in the room but if it’s sufficiently lower than a normal speaking voice then it can be dealt with. If it’s necessary to use noise reduction software to remove some residual hum or ambient tone then the audio will benefit from our large dynamic range (aka large SNR) between the dialog and the noise floor. Noise reduction software makes educated, algorithmic guesses at what is desired sound and what is undesired sound. While it’s impractical for a room full of production professional to attempt to be literally silent, we should all strive to create the lowest possible noise floor.
In contrast to a normal setting, the ambience level of sound in Microsoft’s anechoic chamber, a room specifically designed to absorb all sound, was measured at a world record-breaking -20.6 decibels. It’s an almost unfathomable concept. A world without perceivable sound, yet there in the numbers lie proof that air molecules are vibrating. In fact, according to Richard Gray, that resting ambience level is “close to the limit of what should be possible to achieve without creating a vacuum – the noise produced by air molecules colliding with each other at room temperature is estimated to be about -24 decibels.” It’s been said that people who stay in Microsoft’s room for too long become disturbed and even sick as they begin to hear things like their own blood flowing through their bodies.
Back to our real world scenario for a brief exercise. We’re in a location for a production shoot, say an office building for an interview. I have turned off the HVAC and we’re far from a busy road. We’re shooting after normal business hours so the staff has left. The room is quiet but not uncomfortably silent. I have placed a supercardioid microphone above the interviewee’s head just out of frame. It has an internal self-noise of 12db, just a tiny bit above normal breathing. This noise won’t be heard in the room, only transmitted through the microphone into my recorder. It’s essentially inaudible relative to the much louder resting ambience of the room, which is sitting about 20x louder. Then, our subject speaks. Gloriously lush tone recreating every nuance of timbre pours through my headphones. Everything is going as planned. Yet several interview questions in something happens. A faint noise that no one deems audible is picked up, amplified through my signal flow and immortalized in digital binary code. A noise 100,000 times greater than what we call silence. It’s a director turning the pages of her notes, clocking in a deceptively loud 50db SPL. A sound so quietly loud that it will be heard simultaneously over AND under the dialog, despite the human voice being much louder. Under, as it’s obviously quieter. Yet over, as its unique timbre and frequency characteristics are different enough from the human voice that they stand out significantly when juxtaposed or overlapped.
The loudness of any given noise is relative. Relative to the individual and their own hearing but more fundamentally relative to its frequency and where the perceived pitch falls within our spectrum of hearing, which is 20Hz - 20kHz in a young, healthy person. Not all frequencies are created equally though (to our ears at least). For instance, our ears are more sensitive to “mid-range” sounds, falling between 1kHz-8kHz. This is an evolutionary trait often thought of as “fight or flight” sounds such as babies crying, twigs breaking, various animal sounds, etc. Lower pitch sounds require much more energy at our ear drums in order to be perceived as having an equal loudness to their mid-range counterpart. In fact, Equal Loudness is a scientific term to explain this contour in our hearing sensitivity. The graph to the right is an example of a Fletcher-Munson curve, which is one of many equal loudness contours to explain this phenomenon. The take away for this is that if you absolutely must make a sound on set, make it a very low frequency sound.
Now, back to our “offending” paper turning sound over the spoken word. This of course is a normal occurrence that is easily remedied with a simple polite explanation. No harm, no foul. However, it’s entirely possible that the paper turning has essentially ruined the audio for that particular take….and that may just be the take that you want for the final cut. What then?
So far, we’ve dealt with how loud a sound is perceived in the real world. What we’re really concerned with is how loud any given sound is after it’s recorded. Production sound mixers carry an arsenal of tools that help baffle offended noises (sound blankets) and reject the recording of sounds off camera (highly directional microphones). However, it’s often times impossible to not record a sound that is happening simultaneously to what you’re really after. Can it be “fixed in post”? Perhaps. Let’s take a look at some visual examples of the paper turning in question. Below is a spectrogram of an interviewee responding to a question. The director of this interview was 4.5 feet from the subject sitting just left of camera. The microphone for this interview was a Sennheiser MKH-50 Supercardioid boom mic placed roughly 9 inches from the subject’s mouth. A typical situation for a fairly tight picture frame. To the right of this image is the isolated sound of pages being turned.
The spectrogram graphs amplitude intensity from blue (quiet), black (literal silence) to orange. As you can see, the voice is significantly louder than the paper. However, it’s obvious that the page turns are noticeably audible, as indicated by the orange color on the spectrogram. The important take away here is to notice where on the frequency spectrum the offending noise lies. Notice how it overlaps with where the voice is?See that faint horizontal line running throughout the clip at around 19kHz? hat’s likely the buzz of a light or something coming through the AC electric. While it’s inaudible due to both it’s amplitude and it’s place near the end of the range of human hearing, it should still be removed. Luckily, a sound like that is easily eliminated.
There’s no part of the human voice that will be in this range in any kind of practical way. The next image shows the paper superimposed over the voice. Can you tell which is which from looking? I can’t. While it’s possible to remove specific transients of the offending noise, it’s virtually impossible to remove every bit of it. You could set EQ notches to remove the noise, which is in the upper mid-range. However, then you’d be left with a voice that sounds like it’s underneath a pile of blankets and coming from another room. Thus, unusable. You can set all the noise reduction plugins to their maximum intensity but then you’ll be left with a voice that’s riddled with digital artifacts and sounding like it’s underwater and recorded with a potato. It would seem like the only option then is to not make any noises!
This thought exercise is really just meant to aid in a mental framework where one considers every movement as an action that causes sound. Sound that may not be desirable in a recording. Although it’s impossible for a room full of production professionals to be truly silent while working, I still advocate that people approach this line of work with a degree of empathy. Not empathy towards how your sound mixer feels. Not empathy in any personalized sense that requires an emotional investment. I’m speaking about a degree of empathy towards the shared common goal of the work. Work that requires a commitment towards silence, however far from that goal we reach so that we can all reach that glorious land of aural euphoria. There’s probably some kind of zen-like proverb in here somewhere about the mindful intent being the act itself.
Why all this information? What use will this information provide? Put simply, I hope to give you a new perspective on how loud you are. Now shut up and let me record!
If a crew member makes a noise and there’s isn’t anyone around to hear it, will my microphones still pick it up?
Yes, yes they will.