Frame-accurate placement
Every sound is muxed at its own timestamp with millisecond precision, not nudged into an approximate spot.
Upload a clip. We spot what should be heard and when, generate each sound from the objects on screen, and lay it on the exact frame.
Free to try, no sign-up — 6 sound takes on the house.
Untitled project
No clip loaded
Import a video to start spotting Foley on the timeline
Drop a video clip here
or browse for a file
MP4, MOV, WebM or AVI · up to 25 MB · up to 10 seconds. Clips of 3-15 seconds with one clear action give the best results.
Four steps from a silent clip to a mixed soundtrack.
Drop in an MP4, MOV, WebM or AVI. No account needed to try it.
Surfaces and objects are read from the footage, then every moment that needs a sound is listed with an exact timestamp.
Each moment gets its own audio file from a dry recipe: material, action and surface — not a trailer whoosh.
Audition alternative takes, rewrite descriptions, nudge timings, set levels, then re-render.
Automatic sound design is only useful if you can fix what it gets wrong.
Every sound is muxed at its own timestamp with millisecond precision, not nudged into an approximate spot.
Do not like a hit? Generate another take and switch between them. Switching costs nothing.
The AI's plan is a starting point. Timing, length, level, wording and which sounds are used are all yours.
A tiled bathroom and an open field get different sound treatment, because the palette is decided before any sound is written.
Mixed video, audio on its own, a single sound, or every sound as named stems for your editing timeline.
Reasoning streams to your screen while it works, so a slow job is transparent rather than a spinner.
Give a punch, a swipe or a footstep real weight without digging through a sound library.
Silent animation renders get Foley that lands on the frame the action happens.
Prototype UI and gameplay sound against real captured footage.
Use it as a first pass, download the stems and finish in your DAW.
It watches your clip at five frames per second, lists the moments that need sound with a timestamp for each, generates an audio file per moment, then places every sound at its exact time and mixes the result onto the original video.
Yes. The home page lets you upload a clip up to 10 seconds and generate sounds straight away, with no sign-up. Creating a free account raises the monthly allowance and saves your projects to a library.
Before writing individual sounds the model decides a scene palette: the setting, room acoustics and recording style. Every sound prompt is composed against that palette, so a tiled kitchen and an open field do not end up with the same generic thud.
Yes. Every moment can hold several sounds. Generate another take, audition them, layer more than one on the same timestamp, and re-mix. Switching between takes you already generated is free.
Everything is editable. Change the timing, the length, the level, rewrite the sound description, turn individual sounds off, or keep the original audio underneath the new effects.
The mixed video, the mixed audio on its own, any single sound, or every generated sound as a stems archive named after each event so it drops straight into an editing timeline.
Commercial use is included on the Creator and Studio plans. The free and guest tiers are for evaluation and personal projects.
Start with the free tier and upgrade only when you need more takes, longer clips and commercial rights.