What Reels Translation Does to Your Video.

Instagram now translates Reels audio and captions into other languages, and can move the creator’s mouth to match the new words.

Adam Mosseri announced the latest expansion in July, adding Japanese, Korean, French, German and Italian to a list that already covered English, Spanish, Portuguese, Hindi and a run of Indian languages added through late 2025.

For a brand sitting on two years of Reels, that reads like free distribution. Every video you have already made becomes a candidate for an audience you never shot for, with no new production and no media spend.

The catch is that it only works on video that was built to survive the trip. Most brand video was not.

What actually changed

The tool lives in the Reels composer. You pick up to two languages per translation, Meta AI dubs the audio in a voice built to resemble your own, translates the captions, and optionally syncs your mouth movements to the translated speech.

Nothing publishes behind your back. You can review the result, switch the lip sync off, or discard the translation entirely before the Reel goes out.

Translated versions carry a “Translated with Meta AI” label, and viewers who would rather hear the original can turn translation off in their own settings.

The rollout state is the part worth pinning down

Two numbers are in circulation and they do not agree. Mosseri put the total at 14 languages; a separate Meta release put it at 18. Neither side has explained the gap, so treat any specific count as soft and check the composer for the languages you actually care about.

Access is gated. The translation tools sit with creators who have more than 1,000 followers, on Pages or professional accounts, and only in countries where Meta AI is available.

It has also been rolling out in stages since 2025 rather than switching on for everyone at once. So the honest position for a marketing lead is this: some of your accounts have it, some do not, and the only reliable check is opening the composer on each one.

This is a distribution decision, not a settings one

The temptation is to treat translation as a toggle. Somebody finds it, turns it on, and a fortnight later the brand has Italian and German versions of forty Reels that nobody planned, briefed or read.

That is not entering a market. That is publishing in a language your team cannot check, to an audience you have no idea how to answer in comments, with a label on the post telling viewers a machine did it.

The decision that matters sits earlier. It is whether your video was ever language independent in the first place, and that is settled at production, not in the composer.

What does not survive the trip

The audio translates. Almost everything else that carries meaning in a typical brand Reel does not.

• Burned-in English text. Hooks, subtitles, price callouts and end cards baked into the frame stay in English while the voice speaks Italian.
• Wordplay, puns and idioms. A pun is a sound joke. It does not have a translation, it has a replacement, and there is nobody in the loop writing one.
• Jokes carried by delivery. Timing, a pause, a raised eyebrow against a specific word. Translated speech does not land on the same beats.
• Cultural shorthand. A reference to a UK supermarket, a domestic TV format or a local price point reads as noise to a viewer in Seoul.
• On-screen currency, dates and sizing. Nothing converts these, and a wrong-looking number reads as a wrong-looking brand.

Look at that list and you have a fair test of how much of your library is genuinely portable. For a lot of brands it is the product demonstrations and almost nothing else.

Lip sync does not hide a weak script

The intuitive read is that syncing the mouth smooths over the rough edges. It does the opposite.

Sync raises the apparent intent of the video. A viewer who can see that the mouth matches assumes the whole thing was made for them, so the flat script, the reference they do not get and the English text still sitting in the corner all get read as choices rather than artefacts.

An obviously dubbed video buys forgiveness. A convincingly dubbed one spends it. Meta’s own guidance is that results vary and the sync can look slightly off, which is exactly the middle ground where a viewer notices the effort and then notices the gaps.

The back catalogue is the real prize, if it qualifies

Set aside the new stuff for a moment. The commercial case here is the archive, because the marginal cost of a second language on a video that already exists and already performed is close to nothing.

Suppose a publisher has thirty Reels that outperformed everything else in the last year. If those thirty are visually self-explanatory, tightly scripted and free of baked-in English, they are thirty pieces of proven content pointed at a new market for the price of a review pass.

That is also why this belongs in the same conversation as what Instagram counts as original content now. A translated version of your own Reel is your own work, and it is the packaging around it, not the language, that decides whether it earns distribution.

Building video that survives translation

None of this is a post-production fix. It is a production standard, applied months before the video exists.

1. Put every hook and key line in the spoken audio, not only in burned-in text.
2. Use the platform’s caption layer for text where you can, so it has something to translate.
3. Write literally where the meaning matters. Save wordplay for the domestic-only cuts.
4. Shoot the demonstration so it reads with the sound off. If the visual carries the point, the language is decoration.
5. Keep numbers, dates and currency out of the frame, or accept that those Reels are domestic.
6. Brief editors to build the English text version and a clean text-free master in the same session.

Teams that already run to a spec find this a small change. Teams that brief each video from scratch will find it is the whole video production and editing process that needs the standard, not the individual Reel.

What good looks like

A named owner decides which markets are worth entering, before anything is translated. Two languages chosen deliberately beats five chosen because they appeared in a menu.

A qualification pass runs over the catalogue against the list above, and the videos that fail it stay domestic rather than getting shipped anyway.

Someone who speaks the language reviews each translation before it publishes, and someone is on the hook for the comments that come back in it. If neither of those people exists, the honest answer is that the market is not open yet.

How NBK thinks about translated video

New platform features almost never fail on the feature. They fail on the process behind them, and translation is a clean example: the tool works, and the constraint is a production standard nobody wrote down.

NBK’s bias is to fix the standard first. Decide what a Reel must do to be portable, put that in the brief, and translation becomes a switch you flick with confidence rather than a gamble on what the archive happens to contain.

If you are not sure which of those things your process already handles, the social media audit checklist is a reasonable place to start marking your own work.

Next step

If your social output feels busy but not effective, and you are about to point it at a second language, start with an audit. NBK can help find the constraint in the system before you multiply it across markets.