Cues as data
A caption is a text, a start, an end and a safe area. Changing the language rewrites the cues; it does not reopen a timeline, and it cannot clip a line because the box was sized for English.
Voiceover and captions
Watch it muted without losing the story
Caption cues stored as data per language, plus an uploaded or generated voice track.
Planned
Not built. Captions are timed in the scene plan; no text has been rendered into a frame yet.
Most property films are watched muted on a phone. Captions are therefore not an accessibility afterthought bolted on at export; they are part of the scene plan, which is what makes one edit produce six languages without re-rendering the pixels.
How it is meant to work
Voice is optional and licensed
An uploaded narration always works. Generated voice is offered only once the language quality and the voice licence are verified — a synthesised French accent that embarrasses the agency is worse than no voice.
Thy, Arabic and Latin in one renderer
Thai needs different line-height and Arabic needs bidirectional shaping. Both are constraints on the renderer, not a font toggle at the end.
Specification
- Caption formats
- Burned-in, or sidecar SRT per language
- Languages
- EN, FR, ES, AR, TH, ZH-HK — same six as the interface
- Voice
- Uploaded narration at launch; generated voice only after a licence check
- Status
- Phase 3 — not available during early access
Questions
Will captions be editable after export?
That is the intent: burned-in captions for the file you publish, plus a sidecar SRT you can correct and re-upload. Only the burned-in version exists once a film is exported today — which is also nothing, because export is not built.
Next
Start with what exists
The workspace, the roles, the credit ledger and the invitation flow are live. Create the account now and be first in line when this milestone opens — no card is collected.