Audio transient detection finds the clap in a 50 ms window and refines it to 5 ms; visual motion detection catches the slate close. When both agree, confidence rises.
The offset is computed in whole frames and both claps land on the same record frame. In our reference run: spread 0 frames.
You get sync-plan.json for your pipeline and build_sync.py that builds the synced timeline in DaVinci Resolve.
The helper copies and verifies every file before anything is renamed or synced.
Camera and recorder are analyzed independently — audio first, visual agreement when the slate is on camera.
Offset lands on an integer frame; the plan records the clap frame and the spread.
Run the generated script in Resolve (or let your agent run it) to get the synced sequence.
Credits never expire. Sync work is episodic — you should not rent a seat you don't use.
| Plan | Footage | Price | Effective |
|---|---|---|---|
| Free | 1 hour / month | $0 | — |
| Starter pack | 15 hours | $10 | $0.67 / h |
| Shoot pack | 45 hours | $25 | $0.56 / h |
| Season pack | 225 hours | $99 | $0.44 / h |
| Monthly | 25 hours / month | $15 | $0.60 / h |
Compare: desktop tools in this lane sell for $125–249 as a one-time license, and still need you at the machine. Sync here is zero-install, API-first, and builds your Resolve timeline.
No. Detection runs on your Mac beside the files. The cloud stores the job record, the sync plan and the report.
Detection searches the first 15 seconds, and the visual slate motion backs up the audio when the clap is weak. Each result carries a confidence score; below threshold it asks rather than guesses.
DaVinci Resolve today via the generated build script. The plan JSON is format-agnostic, so FCPXML and Premiere outputs are straightforward additions.
No — that is the point. If you have timecode, your NLE already syncs; Clap Sync is for everything shot without it.
One hour a month, no card, local helper required.