agentleFS
yyert123yyert123/miragea-skills/video-dubbing/SKILL.md

video-dubbing

Skill0 starsChanged 2 days agoMITView on GitHub

AI video dubbing and video translation in the creator's own cloned voice, for Claude Code, Codex and Cursor. Translates a video's speech into Spanish, Portuguese, French, German, Hindi, Japanese and more, re-voices every line in your voice, keeps it in sync (each line starts on the original line, gentle speed-up up to 1.2x, drift reported), keeps music and ambience in the pauses, and exports the dubbed MP4, translated SRT captions and a full-length audio track ready for YouTube multi-language audio. The agent writes the translation itself, so no extra LLM or translation key. Use when the user asks to dub a video, translate a video into another language, localize a YouTube channel, make a Spanish/Hindi/Portuguese version, add a multi-language audio track, or wants an open-source alternative to ElevenLabs Dubbing, HeyGen or Rask. Voice and speech recognition by Miragea; stdlib Python + ffmpeg.

  • Reads credentials

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. renky1025/agent-skills★ 11same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository10 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Check MIRAGEA_API_KEY is set (never print it) and ffmpeg -version works.

    line 24
  2. Distinctive:

    keep names, brands, product terms and numbers.

    line 35
  3. Distinctive:

    keep one line per line (do not merge or split lines.

    line 36
  4. Very specific:

    Check for free: python scripts/dub.py check talk.es.dub.json !

    line 37
  5. Distinctive:

    If the report lists lines still running over, shorten just those lines and render again: unchanged lines come from cache and are not billed.

    line 44
  6. Distinctive:

    No lip-sync: best for voiceover, tutorials, talking-head explainers and faceless channels.

    line 60

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Video Dubbing (translate a video in your own voice)

One video in, the same video speaking another language in the creator's own voice. You (the agent) are the translator: you see each line with its time budget and write a translation that fits. The script does the precise work: sentence-level timing, voicing, sync, mixing, captions, loudness.

When to use

  • "Dub this video into Spanish in my voice."
  • "Make Hindi and Portuguese versions of my YouTube video."
  • "Translate my course lesson / product demo into German, with subtitles."
  • "I want a multi-language audio track for YouTube."

Steps

  1. Check MIRAGEA_API_KEY is set (never print it) and ffmpeg -version works. A voice is needed: MIRAGEA_VOICE_ID (a saved Miragea voice), or MIRAGEA_REF_URL (public URL of a 5-15 s clean clip of the speaker) + MIRAGEA_REF_TEXT (its exact words). Only clone a voice the user owns or has permission to use.
  2. Prepare: python scripts/dub.py prep talk.mp4 --to es Transcribes once (cached in talk.mp4.transcript.json, shared with video-captions and video-to-shorts) and writes talk.es.dub.json: one line per sentence with start, room (seconds until the next line) and max_chars (what fits in that room at a natural pace).
  3. Translate: fill every line's "text" in talk.es.dub.json. Rules:
    • spoken, natural target language (how a native creator would say it), same meaning, same tone;
    • stay under max_chars; when tight, drop filler, not facts;
    • keep names, brands, product terms and numbers; do not add content;
    • keep one line per line (do not merge or split lines; timing depends on it).
  4. Check for free: python scripts/dub.py check talk.es.dub.json ! = too long even at 1.2x: rewrite shorter and re-run check until it exits 0. ~ = will play slightly faster: fine.
  5. Render: python scripts/dub.py render talk.mp4 talk.es.dub.json -o dub_out Options: --burn (burn translated captions), --keep-original (add the original language as a second audio track), --bed music.wav (a no-vocals stem, e.g. from Demucs, instead of the ducked original), --under 0 (mute the original under the dub entirely), --max-speed 1.15.
  6. If the report lists lines still running over, shorten just those lines and render again: unchanged lines come from cache and are not billed.
  7. Report: the files, line count, max drift, and any line you shortened. For more languages, repeat from step 2 with another --to (the transcript is reused for free).

Output (in dub_out/)

  • talk.es.mp4: original picture (stream-copied, no quality loss unless --burn), dubbed audio tagged with its language, -14 LUFS.
  • talk.es.dub.wav: the full mix, exactly the video's length: upload it in YouTube Studio as an extra audio track (Languages > Add audio) so one video serves every language.
  • talk.es.voice.wav: voice only. talk.es.srt: translated captions on the dubbed timing.
  • talk.es.manifest.json: per line start, speed, drift and overflow.

Honest limits

  • One voice per video. For interviews with several speakers, dub one speaker per run (edit the dub file to keep only their lines) or use a voice per speaker in separate renders.
  • No lip-sync: best for voiceover, tutorials, talking-head explainers and faceless channels.
  • Without a no-vocals stem, the original speech is ducked (not removed) under the dub; music and ambience play at full level in the pauses. Pass --bed with a separated stem for a clean result.
  • Speech recognition is cloud-based (Miragea); only a 16 kHz mono audio track is uploaded, never the video.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.