Meta AI Now Dubs and Lip-Syncs Reels Across a Growing List of Languages

What Meta shipped for Reels

Meta AI now translates, dubs, and optionally lip-syncs Reels on Facebook and Instagram. The October 9, 2025 announcement added Hindi and Portuguese to an existing bidirectional English–Spanish capability that Meta had expanded in August.

The technical claim is specific: Meta AI "mimics the sound and tone of a creator's voice" so the dubbed result sounds like the same person speaking another language. A separate, creator-enabled lip-sync feature then aligns the translated audio to the creator's mouth movements.

So this is not subtitle overlay or a generic dub track. It is voice cloning plus facial re-timing applied to user-generated video at platform scale.

Read the update timeline as a rollout strategy

The most informative part of the page is its stack of dated updates. The feature launched with English and Spanish, added Hindi and Portuguese in October 2025, then Bengali, Tamil, Telugu, Marathi, and Kannada in January 2026, then Arabic, Bahasa Indonesian, French, Thai, and Vietnamese slated for June 2026.

That cadence — a handful of languages every few months, tracked with precise dates and even a note that additional platforms were added days later — shows Meta treating each language and surface as a discrete deployment rather than a single global switch. The care in dating each expansion suggests these are hard problems being solved incrementally, not a finished pipeline being localized.

Meta frames the language choices around reach: translating Reels is described as "a free and easy way for creators to reach some of our largest Reels markets." The early languages map to large user populations in India, Brazil, and the Spanish- and English-speaking world.

The consent and disclosure design

Meta built three controls into a feature that fabricates a creator's voice and mouth movements. First, eligibility gates: Facebook creators need 1,000 or more followers, while all public Instagram accounts qualify, in countries where Meta AI is available. Second, lip-sync is opt-in per creator. Third, every translated Reel carries a "Translated with Meta AI" label.

Every translated reel is clearly labeled with Translated with Meta AI, so you'll always know when you're watching translated content. You can also choose to turn translations on or off, or to watch reels in their original language.Montana Labs

The viewer-side "Don't translate" option in the three-dot audio and language menu matters. It puts the choice of whether to consume synthetic audio with the viewer, not only the creator — a two-sided consent model for a feature that could otherwise silently alter what someone thinks they are hearing.

The implication: synthetic dubbing is becoming a default distribution layer

For applied teams, the notable thing is not the translation quality claim but the packaging. Meta is normalizing on-platform voice cloning and lip-sync as a free, creator-facing distribution tool, with the label and opt-out treated as standard product furniture rather than special disclosures.

That sets a reference pattern: if you ship a feature that alters a person's likeness or voice, the surrounding controls — eligibility thresholds, per-creator opt-in for the most invasive step, a persistent label, and a viewer-side override — are part of the product, not an afterthought.

The staggered language timeline also signals that authentic-sounding dubbing across dozens of languages is still being earned one release at a time. Meta is publishing the map as it draws it, and the gaps between updates are the honest measure of how much of this problem remains open.