Subtitle style guide template
Subtitling is not transcription. It is a decision, thousands of times over, about what fits in the time a viewer has to read.
What this is about
The moment two people work on the same project, subtitles drift: one breaks lines by grammar, the other by line length; one writes out numbers, the other does not; one labels every speaker, the other only when it changes. Viewers do not notice individual choices, but they feel the inconsistency as friction. A style guide fixes the decisions once so that the work becomes execution rather than judgement — and it pays for itself again with every language version, because a translator who receives the rules produces files that need one review pass instead of three.
A template is the starting point, not the goal: once it has been filled in by hand twice, it is worth asking whether the document should come out of the plan instead.
What runs differently here
In planning terms, that means:
- Maximum characters per line and maximum number of lines, plus the reading speed target the whole project is measured against
- Line break rules: break at grammatical units, keep an article with its noun, and never leave a preposition stranded at the end of a line
- Minimum and maximum duration per subtitle and the minimum gap between two, so cards do not flash or overstay their cut
- Speaker identification and sound description conventions for accessible subtitles, including how off-screen speech and music are marked
- Typography and positioning: font, size, background or outline, safe-area distance, and where a card moves when it would cover burned-in text
Step by step
This order gets you there fastest:
- Decide first who the subtitles are for: translation for a hearing audience and accessible subtitles for deaf viewers need different rules.
- Set characters per line and reading speed as numbers, then test them on the fastest dialogue passage in the film before adopting them.
- Write the break rules with two examples each, one correct and one wrong, because a rule without an example gets interpreted.
- Define timing: how far a card may run past a cut, the minimum gap, and what happens across a scene change.
- Fix speaker labels and sound descriptions with a fixed vocabulary, so the same door is not slamming, banging and closing in one film.
- Specify typography and the safe area against the delivery spec, and check the positioning rule against any burned-in graphics.
- Give the guide to every translator with the video and the source file, and review the first two minutes of each language before the rest is produced.
Common pitfalls
What most often goes wrong in practice:
- Subtitling everything that is said, which produces cards nobody can finish reading and a film watched at the bottom of the frame
- Cards that run past the cut into the next shot, which reads as a technical fault even to viewers who cannot say why
- Accessible subtitles treated as normal subtitles with a label added, leaving music, off-screen speech and relevant sounds undescribed
With TillyGen
TillyGen carries the language and accessibility versions agreed in the briefing through to the delivery list, so the style guide is written for a known set of files instead of for one that keeps growing.
Change one constraint and the consequences travel through the whole plan: affected shots are flagged, the call sheet is regenerated, and nobody keeps working from yesterday's version.
Frequently asked
How is an accessible subtitle different from a translation subtitle?
Accessible subtitles serve viewers who cannot hear the track, so they identify speakers, describe relevant sounds and mark music. Translation subtitles assume the audience hears everything and only need the language carried over.
Should the guide prescribe exact numbers?
Yes, and test them on your own material first. A rule that says keep it readable produces one style per person; a rule with a character count and a reading speed produces one style per project.
How long does it take to get started with TillyGen?
A first project takes under an hour to set up. There is no configuration phase in which templates and fields have to be defined before the tool produces anything.
Can the results be exported?
Yes — as PDF for the crew, CSV for downstream systems and through the API for anything automated. The plan stays the source; the exports are views of it.