Script-anchored captions and dual text
Captions that follow the words instead of the clock: how takes, alignment and dual text keep subtitles correct through re-records.
The problem with timecoded captions
A timecoded subtitle file is correct until the audio changes. Re-record one line and every caption after it is offset — which is why weekly formats quietly rot.
Words as the coordinate system
Hypit aligns each take to word positions, then resolves captions from that map. Change a take and the captions move with it; no retiming, no re-export.
Dual text
A line can carry two forms: what is spoken and what is displayed. Use it for pronunciation, brand names, translated captions or tighter on-screen wording.
Step by step
1. Write segments, not timings
Divide the script into segments and keep one or more takes per segment. You never type a timestamp.
2. Mark what should differ on screen
Write dual text where the spoken and displayed versions should differ, e.g.
<D | Dee>so captions read the way you want them read.3. Bind the caption track
The caption track renders from the script, so styling comes from the stylesheet rather than from per-line overrides.
4. Re-record freely
Swap a take and re-run. Alignment rebuilds, captions re-position, and everything anchored to selections follows.
After this
Once captions are script-anchored, the caption track becomes a design decision rather than a post-production chore.
Keep reading
FAQ
Do I need exact transcripts?
No. Alignment is per take; you provide the script and the delivery, not a manual transcript.
Can captions be styled per word?
Yes — because selections are named ranges of words, emphasis styling can be anchored to the same ranges that drive b-roll and effects.
How does this help multi-language work?
Localised captions come from a localised script; the pacing and styling stay in the stylesheet, so variants do not drift.