Create a dub

01.Create a dub

curl -X POST https://api.thefamiliarlab.com/v1/dubs \
  -H "Authorization: Bearer $FAMILIAR_API_KEY" \
  -F "file=@upload.mp4" \
  -F "target_langs=es,pt,ja" \
  -F "creator_id=cr_8f2k1q" \
  -F "context=Maya, our head of product, walks through the spring release" \
  -F "style=human" \
  -F "review=manual" \
  -F "metadata.title=Spring release: everything new" \
  -F "metadata.description=Full walkthrough of the spring release. Chapters below."

02.The response

{
  "dub_id": "dub_31xk9m",
  "name": null,
  "status": "queued",
  "source_lang": null,
  "target_langs": ["es", "pt", "ja"],
  "style": "human",
  "review": "manual",
  "outputs": [
    { "lang": "es", "status": "queued" },
    { "lang": "pt", "status": "queued" },
    { "lang": "ja", "status": "queued" }
  ],
  "created_at": "2026-08-13T17:02:11Z"
}

03.Context

contextis the "What's this video?" one-sentencer from the console: the names of the people on screen and what kind of video it is. It sharpens translation the way a human translator's briefing would.

-F "context=Maya, our head of product, walks through the spring release"

The field on this request is the job scope -- one video only. Persistent context (per creator, per recurring stream) is written once through Context and applies to every dub automatically; narrower scopes win.

04.Style

style tells the pipeline how to treat the on-screen presence.

05.Review

review=auto (the default) renders straight through. review=manual pauses the dub at in_review after translation: GET the transcript, PATCH any line, POST /render when approved. Credits only spend at render. The full flow is Transcript & review.

06.Read a dub

{
  "dub_id": "dub_31xk9m",
  "name": "episode-42",
  "status": "rendering",
  "source_lang": "en",
  "style": "human",
  "review": "manual",
  "outputs": [
    { "lang": "es", "status": "complete" },
    { "lang": "pt", "status": "rendering" },
    { "lang": "ja", "status": "queued" }
  ],
  "created_at": "2026-08-13T17:02:11Z"
}

The top-level status follows the lifecycle table; each language also reports its own, so you can publish Spanish while Japanese still renders.

07.List dubs

curl "https://api.thefamiliarlab.com/v1/dubs?limit=25&starting_after=dub_2ww10d" \
  -H "Authorization: Bearer $FAMILIAR_API_KEY"

{
  "data": [ { "dub_id": "dub_31xk9m", "status": "complete", … } ],
  "has_more": true
}

Newest first. Page until has_more is false.

08.Delete a dub

{ "dub_id": "dub_31xk9m", "deleted": true }

Deletes the job and its outputs. A dub that is actively rendering can't be deleted mid-flight (409); spent credits aren't refunded.

09.Credits and limits

Dubs draw on your contract's included credits, metered per finished minute per language at your contract's rates (full and audio-only dubs meter differently); past them, overage rides your monthly invoice. Studio jobs never block on balance -- current balance and cycle are at Usage.

Operational defaults: 10 concurrent renders, 10 GB / 24 hours per upload, 100 requests a minute. The full table + the raise path are in Errors & limits.

10.Access