Craft·June 2026

Sound Design Is Half the Edit

Nobody has ever emailed me to say "great room tone."

We've produced well over a thousand videos at With Media, and I have received feedback on almost everything. The color grade. The captions font. The pacing of a hook. One client once had notes on the way a zoom decelerated — which, honestly, I respected. But in all that time, not one person has ever written to say the ambience under a talking-head section made the room feel real, or that the dialogue was so clean they forgot they were wearing earphones on a train.

And yet I can tell you, with complete confidence, that sound is the thing they were responding to the entire time. When a client says a cut feels "premium" and can't explain why, it's usually the mix. When they say a draft feels "off" and can't explain why, it's almost always the mix. Sound is the layer of the edit that nobody sees, nobody names, and everybody feels.

That's what this essay is about. Not the gear, not the plugins — the thinking. Why half of what people call "cinematic" is happening in their ears, and why it took me embarrassingly long to admit it.

Eyes forgive. Ears don't.

Here's a small experiment you've already run a hundred times without noticing. You're scrolling, and a video opens on a slightly shaky shot. Maybe the white balance is a touch warm. Maybe the framing is a little loose. Do you leave? Almost never. Your brain quietly corrects for it and moves on. Visual imperfection reads as human. Sometimes it even reads as authentic — half the biggest creators in the world shoot on phones in bad light on purpose.

Now imagine the same video, but the voice is thin and roomy, like it was recorded across the hall. Or the levels jump between cuts, so you're riding your volume button. Or there's a faint hiss under everything that you can't quite place.

You're gone in two seconds. And here's the strange part — if someone asked you why you swiped away, you wouldn't say "the audio was bad." You'd say the video felt cheap. Or amateur. Or just... off. You wouldn't be able to point at the problem, because the problem wasn't in the frame.

Your eyes will forgive a shaky shot all day. Your ears flag "cheap" in under two seconds, and they never tell you why.

This asymmetry is the single most useful thing I know about perceived quality in video. Eyes forgive; ears don't. Bad visuals read as honest. Bad sound reads as careless. And carelessness is contagious — the moment a viewer's brain flags the audio as rushed, it assumes everything else was rushed too. The script, the ideas, the person talking. Thin audio doesn't just hurt the video. It quietly discredits the human in it.

The reverse is also true, and this is where it gets interesting for anyone making content. Clean, layered, deliberate sound signals care — before a single word has landed. It buys you trust in the opening second, and trust in the opening second is the whole game in short-form.

The layer I undervalued longest

I'll be honest about my own history here, because I didn't arrive at this by wisdom. I arrived at it by being wrong for a long time.

When I started editing, sound was the thing I did last, in the least time, with the least thought. Drop a trending track under the cut. Slap a whoosh on the transitions. Normalize the dialogue if I remembered. Export. In my head, editing was the visual work — the cuts, the color, the motion. Audio was the plastic wrap you put on at the end.

I've hired somewhere near a hundred editors building With Media, and I can tell you this isn't a me problem. It's the default setting of almost every editor I've ever interviewed. Show me a portfolio and I'll see beautiful grades, clever transitions, kinetic captions — and a mix that's just music at minus twelve and dialogue at minus six, coexisting like flatmates who don't talk.

Why does this happen? Partly because sound is invisible in a portfolio thumbnail. You can screenshot a grade. You can't screenshot a mix. Editors optimize for what can be seen at a glance, because that's what gets them hired. And partly because nobody teaches it. Every editing tutorial on the internet is about what's on the screen. The audio ones exist, but they're filed under a different profession — "sound design" — as if it were someone else's job, in some other building.

The shift for me came slowly, and then all at once. The slow part was noticing a pattern in our own work: the drafts clients called "expensive-looking" weren't always our best-shot pieces. Sometimes the footage was ordinary. But the mixes on those pieces were dense — cleaned dialogue, real ambience, designed accents, music that moved with the cut. The fast part was the day I watched one of our own videos muted and realized it fell apart. Not visually — visually it was fine. It fell apart the way a joke falls apart when you explain it. All the momentum, all the feel, had been living in the audio the whole time.

I run a studio, so yes, obviously biased toward "hire people who obsess over this." But the bias came after the lesson, not before.

What a real mix is actually made of

Let me get concrete, because "good sound" is a uselessly vague instruction. When we finish a piece at With Media, the audio isn't one track. It's a stack of layers, each doing a different job:

five layers most feeds collapse into two.

Clean dialogue. The foundation. Noise removed, plosives tamed, levels consistent across every cut, intelligible on the worst phone speaker your viewer owns.
Ambience. The continuous bed of the space — room tone, distant traffic, a kitchen's low hum. The layer nobody hears and everybody misses.
Design hits. Whooshes, risers, impacts, small accents that punctuate cuts and steer attention. The grammar of the edit, spoken out loud.
Detail sound — the ASMR layer. The sizzle, the pour, the knife on the board, the click of a lid. In food, product and process content, these aren't decoration. They're the content.
Music. Chosen for the piece and cut to the edit — its swells and drops landing on your visual beats, not just playing underneath like a radio someone left on.

Five layers. Most videos on your feed have two: dialogue that was never cleaned, and music that was never cut. That's the gap. That's most of the gap, honestly — the difference between "fine" and "how does this feel so expensive" is usually three missing layers of audio, not a camera.

A finished mix is also about relationships between the layers, not just their presence. The music has to duck — genuinely duck, shaped by hand, not by a compressor preset — every time a voice enters. The ambience has to sit low enough that you'd never name it and high enough that its absence would feel like a dropped call. The design hits have to be rare enough to mean something. An edit where every cut whooshes is an edit where no cut whooshes.

Room tone, or why "clean" silence feels wrong

Of all the layers, room tone is the one I most enjoy explaining to clients, because it sounds like a scam until you hear the difference.

Room tone is the sound of a room doing nothing. The air conditioning you've stopped noticing. The building's electrical hum. The specific way a space with curtains sounds different from a space with glass. Every room on earth has a sonic fingerprint, and your brain — without your permission — is checking for it constantly.

Now, here's what happens in a typical amateur edit. The editor runs noise reduction to clean up the dialogue, and it works a little too well. Between sentences, the audio drops to true digital silence. Absolute zero. And your brain, which has spent your entire life in rooms that sound like something, immediately notices that this room sounds like nothing.

We call that a vacuum-sealed cut. The person is talking, and between their words, the world disappears. It reads as artificial in a way most viewers can't articulate — the same way a photo with over-smoothed skin reads as fake even when you can't point to the flaw. Real spaces breathe. A cut with no room tone is a person speaking in a void, and voids make people uneasy.

So we do the counterintuitive thing: after cleaning the dialogue, we put noise back in. A consistent bed of the actual room, recorded on set — thirty seconds of everyone standing still, which is the strangest half-minute of any shoot day — laid under the entire scene, smoothing over every cut. Now when the sentence ends, the room continues. The space stays alive. Jump cuts stop feeling like jumps, because the ambience carries you across them like a handrail.

Real rooms are never silent. When your edit is, the viewer's brain notices the world has been switched off — even if the viewer never does.

Is this fussy? Enormously. Does anyone consciously perceive it? Never once. Would you feel it missing within a second? Every single time. That's room tone in a nutshell, and it's a decent one-line definition of sound design in general: the craft of things that are only noticeable when they're absent.

What food content taught us

The place where all of this stopped being theory for me was food.

When we started working with Dr. Shilpa Arora, her account had around three thousand followers — most of them bots. Over the run that took her to a million on Instagram and past seven hundred thousand on YouTube, we produced an enormous amount of food content. Morning Routine. What I Eat In A Day. The Ghee Series. The Banana Smoothie video. Recipe after recipe, kitchen after kitchen.

And food content teaches you a lesson about sound that no talking-head video ever will: in food, sound is appetite.

Think about what actually makes you hungry in a cooking video. It isn't the wide shot of the kitchen. It isn't even the close-up of the dish, not really — you've seen a thousand dishes. It's the sizzle. The specific crackle of something hitting hot ghee. The glug of a pour. The clean percussion of a knife going through something crisp. Those sounds bypass the rational brain entirely and go straight to whatever ancient part of you decides it's time to eat. A cooking video with weak sound is a menu. A cooking video with great sound is a smell.

The problem is that a kitchen, acoustically, is a war zone. The exhaust fan drones. The fridge hums. Pans clatter off-camera. If you just use the camera audio, your beautiful sizzle is buried under a wash of noise, at the same distance as everything else, emotionally flat.

So on the Shilpa food content, we stopped treating those sounds as things the camera happened to catch and started treating them as characters. The sizzle, the pour, the chop — each one isolated, cleaned, and then mixed to be felt: pushed forward, given weight, placed exactly on its visual moment so the crack of the tadka lands the instant you see it. Sometimes the best take of a sound and the best take of a shot aren't the same take, and you marry them in the edit. Nobody watching knows. Everybody watching feels it.

Did this move the numbers by itself? I can't isolate it and I won't pretend to. Growth like that run is a hundred decisions compounding — strategy, consistency, her genuine authority on the subject, thumbnails, hooks, everything. But I know what those videos felt like before the sound pass and after, and the difference wasn't subtle. Before: informative. After: you could almost smell the ghee.

Mixing a video to be looped

Which brings me to the video that's sitting just below this essay — "40g Pure Veg Protein in 20 Mins."

On paper it's a recipe video with a promise in the title. In the edit, we made a specific bet: this piece would be mixed for the loop, not just the first watch. Short-form lives or dies on the replay. When a reel loops and the viewer doesn't swipe — watches it again, half-hypnotized — the watch time doubles, and watch time is the metric everything else bows to. So the question in the mix session wasn't only "does this sound good?" It was "does this sound rewatchable?"

Those are different questions. A first watch is carried by information — what is he saying, what's the recipe, where is this going. A second watch is carried by texture. The viewer already knows what happens. What keeps them there is how it feels to sit inside the video again. And texture, in a recipe video, is almost entirely audio.

So every sizzle and pour in that piece was treated like a beat in a song. Isolated and cleaned, yes — that's table stakes by now. But also placed: spaced through the edit so the piece has a sonic rhythm independent of the voiceover, each one arriving with enough weight to be a small reward. The pour lands and it's satisfying the way a drum fill is satisfying. The sizzle opens up and you get that little involuntary exhale. By the time the video ends and loops back to the top, the sounds aren't information anymore. They're the hook. You stay for one more sizzle the way you stay for one more chorus.

Retention isn't only visual pacing. There's an audio pacing underneath it, and the loop is won or lost there.

Watch it below with the sound on, then once muted. The muted version is a competent recipe video. The version with sound is the one you accidentally watch three times. Same frames. Different video.

Music is an edit decision, not a bed

One more layer deserves its own section, because it's the one people think they've already handled: music.

Most creators choose music the way you'd choose a screensaver — something with the right vibe, laid under the whole cut at a polite volume. And look, that's not wrong. It's just leaving most of the value on the table. Because music isn't a bed. Music is a co-editor with strong opinions about time.

Every track has architecture — intros, builds, drops, breaths, the bar where the drums fall away. When your visual edit ignores that architecture, the viewer feels two clocks ticking out of sync: the rhythm of the cuts and the rhythm of the track, each politely pretending the other doesn't exist. When the edit uses it — the reveal landing on the drop, the cut to silence exactly where the track breathes, the pace of the cuts tightening as the build climbs — the two clocks fuse, and the video acquires that propulsive, inevitable quality people call cinematic without knowing what they're naming.

In practice this means our editors cut music like footage. We chop tracks apart, loop a section because the visual moment needs four more seconds of build, pull a drop earlier because the reveal can't wait, sometimes end a video on the track's breath instead of its climax because the quieter landing suits the content. The track that plays under a finished With Media piece is often a Frankenstein of the original — and it fits the edit like it was scored for it, because effectively it was.

Nobody notices. You see where this essay keeps going, right? Nobody notices any of it. That's the job.

The objections, honestly

A smart reader has been arguing with me for a few sections now, so let me take the arguments head-on.

"Most people watch on phone speakers. Why obsess over nuance they can't hear?" Partly true, and it actually raises the bar rather than lowering it. A phone speaker is a brutal little device — no bass, no stereo width, no mercy. Nuance survives on it only if the mix was built deliberately: dialogue carved out clearly, key sounds pushed forward, nothing important living in frequencies the speaker can't produce. A lazy mix falls apart on a phone speaker worse than on headphones. And the headphone audience — commuters, gym-goers, the late-night scrollers — is enormous and hears everything.

"Plenty of viral videos have terrible audio." Absolutely. Raw moments, lucky captures, a clip so inherently interesting that nothing else matters — sound won't save or sink those, and if you've caught lightning, post the lightning. But virality is a lottery and sound design is compounding interest. If you're building a body of work — a coach publishing weekly, a brand showing up every day — you don't get to rely on lightning. You need every ordinary video to feel a notch above its budget, and sound is the cheapest place that notch lives.

"I can't afford a sound designer." You may not need one, in the specialist sense. What the work needs is an editor who refuses to treat audio as an afterthought — who cleans the dialogue properly, records thirty seconds of room tone, pulls the key detail sounds forward, cuts the music instead of laying it. That's craft and time, not exotic equipment. Which is exactly my point about it being the cheapest upgrade: a better camera costs real money and improves the layer viewers already forgive. A better mix costs attention and improves the layer they never forgive.

And one honest concession: there is a point of diminishing returns. If your dialogue is clean, your levels are consistent, your room breathes, and your music moves with the cut, you have collected most of the prize. The last five percent — the layered risers, the sub-drops, the sound-design flourishes — matters for the work we do because clients hire us for exactly that margin. If you're a creator doing your own edits, get the fundamentals right and forgive yourself the rest.

The half nobody sees

I keep coming back to that first line — nobody has ever emailed me about room tone. It used to feel faintly tragic, all that invisible work. It doesn't anymore.

Because the note does arrive; it just arrives in disguise. It's "this one feels really premium." It's a client watching a draft twice in a row without realizing they've done it. It's a viewer who can't tell you why one recipe video made them hungry and an identical one didn't. The compliment never names the sound, the same way you never compliment a building's foundation. You just live in it, and it holds.

Half your edit is happening where no one is looking. I spent years treating that half as an afterthought, and I watch new editors make the same trade every week — hours on a grade the viewer will forgive, minutes on a mix the viewer never will. If you take one thing from this: tonight, watch your last three videos with your eyes closed. Just listen. You'll know within a minute whether your work sounds like it was cared for.

If it doesn't, the good news is that this is fixable — more fixable than almost anything else about your content. You don't need a new camera, a new city, or a new face. You need someone to treat the invisible half like it's half.

Because it is.

“40g Pure Veg Protein in 20 Mins” — every sizzle and pour mixed for the loop.