Keep the picture unchanged: a hand closes a wireless-earbud case. Replace the sound with a bright click, a short muted thud, or a metallic ring, and the viewer's judgment of weight, hardness, and precision begins to shift. The object has not become heavier. Sound has supplied an invisible structure.
This is also where product-ad sound design is easiest to overdo. Low end, impact, and layered effects can create instant polish. But when the auditory cue disagrees with the visible contact, material, or travel of the mechanism, it stops functioning as product evidence. It becomes an attractive but suspicious stand-in.
Good Foley does more than make an action audible. It makes us believe that this object, at this contact point, under this amount of force, could have produced that sound.
The ear hears physical clues, not a single adjective
Light or heavy, soft or hard, is not a fixed translation of one frequency. An impact contains an onset, amplitude, frequency distribution, resonances, and decay. Listeners interpret that combination alongside the visible size, material, applied force, and previous experience.
Acoustic research helps define the boundary. In a study of the auditory discrimination of impact force, Lutfi and colleagues found that listeners used amplitude and frequency cues, but did not weight them uniformly. Another study of struck bars related frequency, amplitude, and decay to density, stiffness, damping, length, cross-section, and hollowness, while also showing that some physical differences fall below reliable auditory sensitivity.
A multisensory experiment offers an even more useful warning against treating a metallic sound as a universal shortcut. Liu and Ando compared material categories with contact sounds whose frequency and damping had been modified. Material category affected audio-only judgments, but low-level acoustic parameters—especially damping—interacted with touch in more complicated ways. The practical lesson is not that higher frequency always means harder. It is that the category of a sound, its detailed envelope, and the visible action must be reviewed together.
Marketing research has also found that lower-pitched voices or music can make people infer a larger product, but that the effect is shaped by visualization cues and becomes weaker when the product is visibly present. Foley can reinforce a product judgment; it cannot reliably rescue a persistent conflict between what people see and what they hear.
Find the contact cause before deciding whether a sound is pleasing
A closing action often contains several events: the finger applies force, a latch crosses its resistance point, two surfaces meet, the shell resonates, and an internal part may rebound. The shot may show only a few frames, but the ear still looks for this causal chain.
The sound editor should therefore mark three moments before asking whether the result feels substantial: where the force begins, which frame contains the visible contact, and whether a real resonance remains after the action stops. An onset that arrives too early feels as if the effect is pushing the picture. One that arrives after the motion has finished makes the object seem sluggish. A long ring can turn light plastic into an imaginary metal cavity; removing every tail can flatten real metal or glass.
This is different from the spatial continuity discussed in ONCE's article on ambient sound in brand films. Ambience keeps a room alive across an edit. Product Foley solves a smaller causal question: which part touched what, at what moment, and with what kind of resistance?
Three sound layers are safer than one luxury effect
The truth layer: record the product's own answer
Recording the real product in a quiet space does not mean the final mix must use the raw take unchanged. The recording is first a ruler. It shows the true travel time, number of contacts, shell resonance, and differences created by microphone angle or table reflections.
Capture several normal operating forces and grips, but do not make the performer exaggerate the action merely to create more sound. If the product is quiet, preserve that fact. Quietness may express controlled damping, precision, or usage etiquette—not a defect waiting to be replaced by a cinematic thump.
The readability layer: use Foley to clarify the true relationship
Production sound can be masked by fingers, clothing, or the room. Foley may use another prop to create a clearer onset, friction, or rebound. The substitute does not need to look like the product, but it must behave like the visible cause.
If the reference contains one latch, one shell contact, and a short resonance, those elements can be recorded separately and recombined. If the picture shows one contact but the soundtrack supplies three mechanical locks, sound has started inventing a structure. Readability should refine evidence, not add parts the product does not have.
The expression layer: let product sound enter the ad's rhythm
Only after the truth and readability layers work should the sound be fitted to music, edit accents, and brand tone. The action can become a rhythmic beat, or occupy half a second of space after music steps aside. Expression should not undo the first two layers. Moving contact several frames to hit a drum, or stretching a tail into metallic resonance to signal expense, loosens the physical relationship.
For a broader system of reusable sonic identity, see ONCE's article on sonic memory in brand films. The question here is narrower: how a single product action can preserve believable material and weight without turning every click into a sonic logo.
AI can produce candidates, but it does not know which sound belongs to this unit
Video-to-audio research in 2026 has made synchronization a much more explicit problem. Sony AI's FoleyBench evaluates visible sound sources, audiovisual alignment, temporal synchronization, and audio quality; its authors report that 74% of videos in previous evaluation datasets had poor audiovisual correspondence. FoleyDirector uses structured temporal scripts to control sounds in multi-event scenes, while Apple's StereoFoley addresses object-aware stereo position and distance.
These systems can expand a candidate set: adding rough sound to a concept test, comparing material directions, or building an initial track for generated footage. But the papers still evaluate semantic, temporal, and spatial alignment separately. Generating something that resembles a closing sound is not the same decision as establishing that this sound belongs to this product.
Critical sounds in live-action product footage should still begin with a real reference. AI or a library can contribute timbre and layers; a sound editor must align them with the actual contact, scale, distance, and context, then verify usage rights. If the result suggests metal, sealing, locking, or performance the product does not possess, the problem is not which tool produced the file. It is the creative judgment that crossed the factual boundary.
One track must survive both headphones and a phone speaker
Studio monitors can reveal deep energy, very short transients, and subtle decay that a phone speaker will reproduce differently. If weight exists only in the lowest frequencies, the mobile version may collapse into a hollow click. If the fix is to brighten every mid-high detail, headphones may turn the same action brittle.
The goal is not identical sound on every device. It is identical causality. On a phone, the listener should still hear the order of contact and rebound; headphones may add space and detail. Neither playback should imply a completely different mechanism. Music, voice, and Foley must also be tested together in the final short edit, because a clean isolated click may be masked by a drum hit or a consonant once the full mix returns.
A three-pass audition: reference, candidates, endpoints
First, play only the real reference against picture. Note the number of contacts, onset frame, travel, and residual resonance without asking whether it feels premium. This establishes the factual floor.
Second, synchronize two or three Foley candidates to the same image and run a blind description test. Ask listeners what material and weight they infer, and whether they hear a latch, cavity, or rebound, before revealing the intended direction. If the description repeatedly disagrees with the real structure, the most satisfying candidate should not automatically win.
Third, return to actual endpoints: studio monitoring, ordinary headphones, and a phone speaker. Play the compressed short-form cut with music and dialogue, then watch it once on mute to ensure the claim does not exist only in audio. The final choice is not the loudest sound across three devices. It is the one that preserves the same contact cause across them.
Product-ad sound design should not disguise a light object as a heavy one or turn every touch into a trailer impact. Its job is to supply the right resistance, contact, and rebound beyond the image—to make the product's structure believable and show that the creative team has paid attention to every physical detail.
If you are planning a product commercial, brand film, or hybrid live-action and AI production, ONCE video production can develop the action, reference recording, Foley, mix, and format adaptations as one credible audiovisual system. Send us your project brief to begin.
Sources and scope
This article reviewed studies in the Journal of the Acoustical Society of America on impact force and the material and geometry of struck objects; Liu and Ando's 2016 experiment on auditory and haptic stiffness; Lowe and Haws's 2017 six-study paper on pitch and inferred product size; and current public research materials for Sony AI FoleyBench, CVPR 2026 FoleyDirector, and Apple StereoFoley. The three-layer model and three-pass audition are ONCE's editorial analysis, not guarantees of advertising performance or product capability.
