Florentyna - Subtitle Translator AI Skill

Turn a dialogue transcript into a subtitle file viewers can actually finish reading before the cut, not just a technically accurate translation on screen.

Florentyna is a Senior Subtitling & Closed-Caption Specialist skill built to treat every caption as a timed object: setting a characters-per-second reading-speed budget before writing a single cue, breaking lines at natural grammatical boundaries rather than mid-thought, syncing every cue's in and out points to shot changes rather than just to speech, deciding SDH versus standard subtitles before drafting, condensing dialogue for meaning rather than word count when it outruns the budget, and running a dedicated QA pass for timing drift, reading-speed violations, and overlap errors before delivery.

Who it's for

video producers, streaming teams, and content creators who need reading-speed-compliant, properly synced subtitle or caption files.

Key capabilities

• Calculates a hard characters-per-second ceiling from cue duration before formatting any line, treating an over-budget cue as broken, not merely tight.
• Breaks two-line subtitles at a clause or conjunction boundary, never mid-phrase or across a name.
• States whether she's building SDH or standard subtitles before drafting, since SDH carries speaker ID and sound-effect content standard subtitles correctly omit.
• Condenses dialogue by cutting filler and restructuring for economy first, trimming meaning only as a last resort and noting when it happens.
• Runs a dedicated drift QA pass across the full runtime, checking timing drift, reading-speed compliance, and overlap or gap errors separately.

How to use it

Paste Florentyna's SKILL.md into your Claude Project Instructions (or any AI system prompt), then describe your subtitle timing or caption problem. Works with Claude, ChatGPT, and any AI chat. Under 2 minutes to install.