Workflow guide
Sound effects that hit hard and sit quietly
Viewers forgive shaky footage faster than bad sound, and the right effects make motion feel physical — punches land, interfaces click, scenes breathe. The failure mode is always excess: every moment emphasized means nothing is. This guide covers spotting, layering, and mixing effects that serve the cut.
From the Mellius product team · Updated September 28, 2026
Conspiracy theorists, you may hate this
AI Director
›
Just type what you need
- ›Remove all silent parts
- ›Add bold kinetic captions
- ›Only have Bill Ackman speaking
- ›Cut to highlights only
- ›Add cinematic transitions
- ›Match cuts to the beat
The short answer
Spot your edit for moments that need emphasis — impacts, whooshes, UI clicks, ambience beds — then layer at most two or three effects per hit, panning and pitching each slightly for width. Keep effects 6 to 12 dB under dialogue, duck music beneath both, and check the final mix on phone speakers before export.
Every sound earns its place
Each effect marks action, transition, or mood — decoration tracks get muted.
Dialogue stays king
Speech remains intelligible on phone speakers with effects and music underneath.
Mix translates everywhere
The balance holds on headphones, laptop speakers, and a phone in a noisy room.
Spot the edit before touching the library
Watch the locked or near-locked cut and mark every moment that wants sound: impacts, whooshes under motion graphics, clicks for UI, risers into reveals, and ambience beds that glue scenes together. Spotting first prevents the classic trap of auditioning effects while the edit still moves, which wastes hours on moments that get cut. Prioritize ruthlessly — a five-minute video rarely needs more than a dozen featured effects plus beds.
Choose recordings over synthesizers when physicality matters: a real door, clap, or footstep layered quietly under a designed hit sells weight that pure synthesis rarely reaches. Keep a favorites folder per project so recurring sounds stay consistent, and note the timecode of each spot so placement survives timeline shifts. In Mellius you can describe where effects belong in words — a hit on the logo impact, a riser into the reveal — and adjust each one on the editable timeline with undo intact.
- Spot on a near-locked cut, noting timecode for each effect.
- Budget roughly a dozen featured effects per five minutes.
- Prefer recorded layers for physical weight, designed layers for sheen.
- Keep project favorites so recurring sounds match.
Layer, pitch, and mix like a record
Build each featured moment from two or three layers — attack, body, and tail — pitching and panning each slightly so the stack sounds wide rather than doubled. A punch might combine a low thump, a mid crack, and a short room tail; nudge layers by a few milliseconds so transients fuse instead of flamming. Trim every effect tight to its picture cue with short fades, since audible heads and tails scream library demo.
Mix against dialogue, not solo: park speech where it sits naturally, then tuck effects 6 to 12 dB beneath it and duck the music bed further still. Check the mix at low volume on phone speakers, where masking problems hide — if the dialogue survives there, it survives everywhere. Common mistake: normalizing each effect to full scale, which leaves no headroom and forces the whole mix into a limiter.
- Stack attack, body, and tail; detune and pan layers slightly.
- Fade every effect head and tail to its picture cue.
- Keep effects 6–12 dB under dialogue, music lower still.
- QC the mix quietly on phone speakers before export.
Compare the workflows
| Source | Best fit | Trade-off |
|---|---|---|
| Built-in editor library | Speed on everyday creator work | Smaller selection, recognizable sounds |
| Recorded foley | Physical, bespoke texture | Needs a quiet room and a decent mic |
| Licensed effects packs | Cinematic depth and variety | Check licenses for client and broadcast use |
A practical sequence
Put it into practice
- 01
Spot the cut
Mark impacts, transitions, UI, and ambience on a near-locked edit with timecodes.
- 02
Layer each moment
Stack two to three pitched, panned layers per hit and trim tight to picture.
- 03
Mix under dialogue
Tuck effects below speech, duck music, and QC quietly on small speakers.
Frequently asked questions
How loud should sound effects be relative to dialogue?
Park dialogue first, then sit featured effects roughly 6 to 12 dB beneath it, with music lower still. Momentary peaks like impacts can kiss dialogue level for a frame or two, but sustained effects above speech destroy intelligibility. Always confirm on phone speakers at low volume.
Where can I get sound effects legally?
Start with your editor's built-in library, then expand to licensed packs with clear client and broadcast terms. Record simple foley yourself for bespoke texture — doors, claps, and footsteps need only a quiet room and a decent mic. Read every license before client work, since free downloads often exclude commercial use.
Why do my sound effects sound cheap?
Usually a single full-scale library file played dry, with audible heads and tails and no layering. Stack two or three pitched layers, trim tight to the picture cue with short fades, and tuck the result into a balanced mix. Context and level do more for realism than any single expensive sample.
Continue reading
Describe the edit. Keep the timeline.
Mellius turns conversational direction into cuts, captions, graphics, motion, color, and sound you can still inspect and change.
Start editing free