Meta AI Reels translations in 9 languages: burn captions with Sume

Meta AI translation for Reels covers nine languages, per Meta's own announcement (read 2026-10-02): it translates, dubs and can lip-sync your voice. It does not give you reviewed on-screen text. To burn your own translated captions into a clip, translate the text yourself, then send it to Sume's video captions as cues with text, start and end. Sume does not translate or dub for you in that call.

What Meta announced

The page says Meta AI supports translation, dubbing and lip-syncing for Reels in nine languages in total. The tool preserves the sound and tone of the creator's voice, and lip-sync is optional.

When burned-in translated captions still help

A platform dub changes the audio. Burned captions are text in the picture, so they travel with the file to any place you post it and you control the wording and the timing. That helps when a brand term must be spelled one exact way, or when the clip goes somewhere without a translation feature.

How to do it with Sume

Get the source timing from video inspect with transcribe: true and sentence segments[]. Translate each segment's text with any tool you trust, keep each segment's start and end, and pass the result as cues to POST /v1/video-captions. Authored cues skip speech-to-text, so the burn uses exactly your words. The source video_url must be a fetchable public HTTPS URL.

The docs list $0.20 per accepted caption job for videos up to 60 seconds under the current fixed estimate; confirm in GET /v1/catalog.

Two honest limits

First, the language field is only a speech-to-text hint. It does not translate and it does not choose a font. Second, the caption docs document Hangul faces in detail, but I found nothing that says the Latin styles carry Devanagari or Bengali glyphs. Run a short test clip in your target script and look at it before you batch. Translations also run longer or shorter than the source, so check that each cue still fits its time slot.