All products

Translate subtitles with AI

Translate a timed caption track while preserving cue boundaries and speaker labels, then check a single-language or bilingual result against source cues.

Translate subtitles with AI from a track whose timings already work. A fluent line is little use if it appears over the next speaker's answer. The lai-translate skill breaks the caption text into manageable chunks, then merges translations back into their original segments.

Decide whether the audience needs the translation alone or both languages. Provide any preferred terms before working through a long interview. The structural checks can catch missing cues and changed timings, but the final test is still whether the lines make sense and can be read on screen.

What it does

  • Keep each translated cue attached to the intended time range.
  • Preserve speaker labels through the translation.
  • Create target-only or bilingual captions.
  • Flag empty or unexpectedly unchanged segments for review.
  • Run a second wording pass when consistency or natural phrasing matters.

Install and use

  1. Install the skill in your agent. Provide an authorised caption file and the official Python caption dependencies for the local helpers. The normal route uses the current agent, while optional batch translation requires a configured provider and key.
  2. Ask, “Translate this interview for English viewers. Keep the original timings and produce a bilingual version for editorial checking.”
  3. Compare several rapid exchanges with the source and read the longest cues at playback speed before delivery.

FAQ

Source

Made by LattifAI. Original skill: lai-translate, published on GitHub under the MIT licence.

View on GitHub