Translate a subtitle file with AI
Work from an existing subtitle file or an authorised local recording. Keep the original cue times while building a translated or bilingual copy for review.
To translate a subtitle file with AI, start with the episode and track you actually intend to use. One character can begin a sentence in one cue and finish it in the next. The subtitle-translator skill reads across that boundary before fitting the translation back into the original timing. A pause or an interruption can carry as much meaning as a word.
Choose the target language and whether the reader should see one language or both. A glossary and saved context keep earlier choices available across episodes. With authorised local video and no usable dialogue subtitles, an optional speech recognition step can create a source transcript for review.
What it does
- Create a bilingual SRT with an agreed language order.
- Correct a recurring term across a series.
- Keep an original overlap between two speakers intact.
- Check that every supported cue has a translation.
- Compare a difficult exchange with the surrounding scene.
Install and use
- Use a file-capable agent and your subtitle source. The optional checker needs Python 3.10 or later; local transcription also needs FFmpeg and faster-whisper.
- Ask “Translate this episode, keep its timestamps and use the character names in my glossary.”
- Review the separate output in a player, especially fast exchanges and uncertain passages.
FAQ
Source
Made by Subtitle Translator contributors. Original skill: subtitle-translator, published on GitHub under the MIT licence.
View on GitHub