Record a Small Foley Test for the Commercial's Key Action
Record a Small Foley Test for the Commercial’s Key Action
A recorded sound can be interesting and still useless. Solo one that you like and play it back with nothing else running and you will learn almost nothing about whether it does any work. The useful information arrives a second later, when you lay that sound against the action it is supposed to serve and hear whether the action now reads as heavier, more tense, or more final than it did before.
That is the whole test. Foley — the practice of performing sound in time with picture rather than assembling it entirely from library recordings — exists because a performed action carries intention that a found effect usually does not. The Foley mixer Ben Cross, describing an armour sound library he worked on, talks about sounds built from combinations of materials rather than hunted down whole, and about recording perspective as a deliberate choice rather than whatever the room happened to give him (A Sound Effect interview). That is one practitioner's first-person account of one project, and his subject was plate metal, not paper, so what transfers here is the approach, not the setup.
Your version of that approach is small enough to run in an afternoon: one action, one listening question, a handful of takes, and a comparison. The result is a provisional expressive test meant to move a treatment conversation off adjectives and onto a decision. It is not the finished sound, and it is not evidence of what the real product does.
Start with one action and one listening question
Pick the single contact or movement the treatment is weakest on: a grip, a release, a slide, a closure, a change in resistance. Then name what the sound is supposed to add.
The picture already supplies a great deal. It shows the hand, the object, the direction, the speed. What it cannot supply is weight, tension, material stiffness, finality. That gap is your job, and it should be the shape of your question.
Weak question: What does opening an envelope sound like? That produces a folder of noises. Strong question: Does the envelope read as under tension, so the card's arrival feels like a release? That produces a test, because two takes could answer it differently and you could point at the one that does.
Here is a test you can run: a plain paper envelope, and a listening question about tension. The action runs a little over three seconds. The hand contacts the envelope around the one-second mark, the flap begins to give near a second and a half, the card slides free somewhere around two and a quarter and clears by two and three-quarters, and the hand stops near the end. Those numbers are approximate and deliberately so — they are a timing sketch, not a measurement — but you should be able to say them out loud before you record anything.
If two actions both matter, run two tests or accept that you have moved from one question to two. A test becomes unreadable when it is trying to answer three things at once.
Choose material for character, not for resemblance
The instinct is to find the object. Resist it. A performed substitute is allowed to be expressive rather than faithful; that is the reason Foley exists as a craft instead of a filing system. Your ear accepts a sound as the sound of an event long before it is a measurement of that event.
So the material question is not "does this look like the thing" but "does this movement carry the character I asked for." Put two or three ordinary candidates through the same gesture and listen. For the envelope, the gauge of the paper is the whole decision: a light sheet and a stiff card stock give the same event two very different weights, and a heavier stock may read as more consequential even though it is less like the paper actually on screen.
Keep the shortlist safe and dull. Envelopes, boxes, lids, corks, zippers, fabric, a mug set down on wood. Nothing that can cut, burn, shock, or startle, and nothing you have to hold under tension near your fingers or your face. If the action on screen is inherently hazardous to reproduce, perform the movement safely and let the sound live in the material rather than in the mechanism.
Run a setup test before you run takes
You need one microphone, one reasonably quiet room, headphones, a recorder, and the action playing muted on a screen. Not the audio through speakers — that will bleed into your take and it will also pull your performance toward syncing to a sound rather than to a movement.
- Put the mic where the action will happen. Start at a distance that matches how close the audience feels to the object, and be prepared to change it once.
- Record twenty or thirty seconds of the room with nothing happening. This is room tone, and it is the reference for every later question about whether a noise is really in your take. Listen on headphones and name what you hear: fan, fridge, traffic, HVAC, keyboard, your own breathing.
- Perform the action once as a level check rather than a take. Peaks should sit comfortably below clipping. If you are hitting the ceiling, move the mic back a little or drop the input gain before you change the performance.
- Listen back. Can you hear the grip, the give, the slide as separate events? If the take is one smear of low rumble, the distance or the room is wrong, and no amount of performance will fix it.
- Leave a few seconds of stillness before and after every take. It gives you room to trim and it makes the room's contribution audible.
A waveform that fills the screen tells you only that something was loud enough to be recorded. It cannot tell you whether contact detail survived or whether the perspective is useful. Look at the picture, not at the drawing.
Once the setup test passes, hold the conditions. Same mic position, same gain, same room. If the air handler comes on between take two and take three, those takes no longer share a room, and any difference you hear between them is now partly the building. Put a tape mark on the floor for the stand so you can find the position again.
Change performance first, then perspective. Not both.
Now record the two performances. Same envelope, same mic position, same gain, same start position, roughly the same duration. Nothing varies but the hand.
Take A — decisive. One firm grip, one pull, the flap gives in a single motion, the card comes out in one move and stops.
Take B — hesitant. The grip tightens, a first tug that does not move the flap, a pause, then a second pull that opens it. The card comes out slower and scrapes.
Label them as you go, so the raw files stay readable without you explaining them later:
env_open_A_decisive_01.wav
env_open_B_hesitant_01.wav
room_tone_quiet_01.wav
Play A and B back to back against the reference, not alone. The comparison is asking whether the action reads as one decision or two, as under tension or not. Two takes can be nearly identical under a spectrogram and completely different in what the action means, and meaning is what you are testing.
Only after that, change the listening perspective. Take the performance with more events to hear — B, in this case — and record it again twice: once with the mic noticeably closer to the action, once noticeably farther. Resist normalising anything before you have listened. If you do match loudness later, do it on copies, and note that raising the level of the far take also raises the room, which is part of what you are trying to hear.
Be honest about what this comparison contains. A repeated performance is never the same performance, and moving the mic changes the geometry between your hand, the object, and the microphone, which changes what you hear while you perform and therefore how you perform. So this is a perspective-dominant comparison, not a perspective-only one. If the differences are larger than distance can account for, say so rather than crediting the room with everything.
env_open_B_hesitant_micnear_01.wav
env_open_B_hesitant_micfar_01.wav
The reason to separate these two passes is simple. Change the material and the distance in the same breath, and if the result improves you will not know whether it was the paper or the room. A test you cannot learn from is not worth the afternoon.
Align the copies and judge the event
Judge onset before you slide anything. Listen to each take as performed, unshifted, because that is the only version in which an early or late first sound is still audible. Alignment comes after the listening, and it exists for presentation: duplicate the raws, place the copies under the reference, and slide each copy so its first audible event lands on the visible contact, so whoever receives them can compare the takes without hunting for the moment. Each take needs its own alignment, because each was performed separately. Once a copy has been slid, its first sound sits on the contact by your hand rather than by the performance, so a copy will always look correct and cannot tell you about onset.
Listen in this order.
Onset. From the raw, unshifted take: does the first sound land on the grip, or does it arrive before the fingers close? An early arrival reads as anticipatory, which is sometimes exactly the weight or speed you want and sometimes makes the action feel soft, as if the paper moved itself. For a question about tension, a sound that begins before the hand does may argue against you.
Continuation. Between the grip and the give, roughly half a second in this sketch, is anything audible? If the answer is a continuous rustle, that rustle is now covering the interval where the audience watches the paper strain. Sometimes you want the cover. Sometimes the near-silence is where the tension actually sits, and the give lands harder for it.
Decay. When the flap opens and the card slides free, does the sound stop when the object stops, or ring on past the moment? A sound that outlasts the picture tends to read as larger and softer; one that stops hard reads as more final.
Masking. You are hearing this take alone. In the finished piece, something else will occupy that stretch of time. Ask what your take would cover if it sat under a line, or under the moment the card's edge appears in frame. A long bright rustle can easily become the loudest and least useful thing in the shot.
Then make the judgment in terms of the action, not the sound. Take B, because the double tug makes the action read as careful rather than decisive, and the brief silence before the give is where the tension lives. That is a reason someone can argue with. Because it sounds better is not.
Keep the preference provisional. At this scale, the test cannot tell you what the final mix will do to it.
What you hand over, and what you have not proved
Give the sound team the raw takes, labeled and unprocessed; the room tone; the aligned copies, clearly separated from the raws; a sentence or two stating which take you prefer and why; and a note saying which takes share conditions, so the comparison stays readable without you in the room. Say plainly that the copies are aligned for convenience and that onset was judged from the raws, so nobody reads the alignment as a claim about timing.
Keep three things apart while you write this up, because the value of the test depends on it. There is the captured sound: what is actually in the file. There is the constructed impression: what you built by choosing a material, performing an action, and sliding it into place. And there is product behaviour: what the real object does in real use, which none of the above establishes.
A director who says the latch reads as heavier now is making a claim exactly the size of the evidence. A director who says this is what the latch sounds like has promised acoustic proof that three seconds of tape cannot deliver, and it will be the first thing to collapse when the real thing gets shot.
Three seconds, two performances, two distances, one room tone, one aligned comparison, and a sentence of judgment. That is not much tape. It is enough to turn an adjective into a decision someone else can respond to.
Frequently asked questions
What makes a small Foley test readable rather than just a folder of noises?
It starts with one action and one listening question tied to what picture cannot supply, such as weight, tension, stiffness, or finality. Two takes should be able to answer the question differently so one can be pointed to and defended.
Should the substitute material look or sound like the real object?
Not necessarily. A performed substitute is allowed to be expressive rather than faithful. The material question is whether the movement carries the character asked for; for an envelope, lighter paper and stiff card stock give the same event different weights, and a heavier stock may read as more consequential even if less like the on-screen paper.
Why play the action muted instead of through speakers while recording?
Speaker audio can bleed into the take and can pull the performance toward syncing to a sound rather than to a movement.
What should change first: performance or microphone perspective?
Change performance first, then perspective, not both at once. If the material and distance change together and the result improves, you will not know whether the paper or the room caused it. Even then, moving the mic changes the geometry, so the comparison is perspective-dominant rather than perspective-only.
What does the test not establish about the real product?
It keeps captured sound, constructed impression, and product behaviour separate. Three seconds of tape can show what is in the file and what was built by material, performance, and alignment; it cannot establish what the real object does in real use.