Google Ads is rolling out AI video dubbing inside Asset Studio, turning one uploaded creative into versions for 33 languages and locales at no extra cost to the advertisers who have it. The feature is genuinely useful and the temptation it creates is genuinely expensive: dubbing removes the cheapest part of localisation and leaves every costly part exactly where it was.
Asset Studio's AI dubbing translates and voices an existing video ad into 33 languages and locales, free for advertisers in the current limited rollout. It is the right tool when the message sits in the voiceover and the visuals carry no language. It is the wrong tool for close-up talking heads, testimonials, price and legal claims, and any market where you cannot review the output. Dub the audio; you still have to localise the offer.
What Google actually shipped
Video Ads Dubbing sits in Asset Studio, the creative workspace inside Google Ads. It takes a video you have already uploaded, translates the script and generates a dubbed audio track, across 33 languages and locales. Reporting through September 2026 describes a limited rollout to selected advertisers, free for those who have access; the capability was first spotted as an internal feature in late 2025, so this is a controlled release of something Google has been testing for close to a year.
The strategic point is smaller than the headline and more useful. Localising video has always been gated by production cost, which meant markets below a revenue threshold got either nothing or an English ad. Removing the voice-recording step lowers that threshold. It does not remove it.
About that 86% figure
Google's promotional material for the feature cites an 86% increase in click-through rate and a 75% reduction in cost per click. Those numbers are circulating in coverage as if they were benchmarks. They are not: they come from the vendor's own marketing for the product, without a published methodology, sample size, market mix or baseline, and the comparison is almost certainly a localised ad against a non-localised one rather than against a professionally localised alternative.
The direction is entirely believable — an ad in the viewer's language should beat one that is not — but do not put the magnitude in a business case. If you want a number for your own planning, run the test described at the end of this article and use your own.
Free dubbing is not free localisation
This is the part that decides whether the feature saves money or quietly loses it. Dubbing addresses one line in the localisation stack. Here is the whole stack, with what Asset Studio does and does not cover.
| Localisation layer | Covered by AI dubbing? | Who still has to do it | Typical effort per market |
|---|---|---|---|
| Spoken voiceover | Yes | Nobody, beyond review | Minutes |
| On-screen text, supers, end cards | No | Your editor or design team, re-rendering the master | A few hours per variant |
| Prices, offers and currency | No | Marketing, per market | Hours, plus approval |
| Legal disclaimers and claims | No | Legal or a local regulatory check | Days in regulated categories |
| Cultural and idiom check | No | A native speaker who knows the category | An hour per script, and non-negotiable |
| Landing page and forms | No | Web team | The largest item on this list |
| Customer handling after the click | No | Sales or support, in that language | Ongoing |
Run down that column honestly and the conclusion is uncomfortable but clarifying: a dubbed ad pointing at an English landing page, with an English form and no one to answer the enquiry in that language, is a more efficient way to waste money than the English ad was. Dubbing lowers the cost of the first ten seconds; every other step is unchanged.
Before dubbing anything, localise one landing page and make sure someone can respond to an enquiry in that language. Then dub. Teams that do it in the other order generate cheap traffic into a funnel that cannot convert it, and conclude the market does not work.
Dub, subtitle, re-voice or reshoot: a decision tree
Four options exist and the choice is mostly determined by where your message lives and how the creative is shot. Work through these in order.
- Is the market worth a localised landing page and a way to answer enquiries in that language? If no, do not localise the ad either. Run English and accept the result, or skip the market.
- Does most of the message sit in on-screen text rather than speech? If yes, dubbing barely helps. You need a re-rendered version with translated text, and dubbing is a secondary step.
- Is a person speaking in close-up for most of the runtime? If yes, prefer subtitles or a locally recorded voiceover over a lip-mismatched dub. Mouth movement that does not match audio is noticeable in exactly the shots most brand films are built around.
- Does the ad make price, performance or regulated claims? If yes, any language version needs local review regardless of how it was produced. Budget for the review before the production saving.
- Is it a testimonial, a customer story or a named individual? If yes, do not dub without the speaker's consent. You are putting words they never said into their own voice.
- Otherwise — product demo, voiceover-led, visuals doing the work, non-regulated claims — dub it. This is the case the feature was built for, and it is a large share of performance video.
| Creative type | Best option | Why |
|---|---|---|
| Voiceover-led product demo | AI dub | No face to mismatch, message is in the narration |
| Founder or presenter to camera | Subtitles, or a local re-record | Lip mismatch is most visible in close-up |
| Customer testimonial | Subtitles | Consent and authenticity outweigh the convenience |
| Text-and-motion social cutdown | Re-render with translated text | Audio is not carrying the message; many viewers are muted anyway |
| Connected TV or YouTube brand film | AI dub plus a native-speaker review | Sound-on environment, high reach, dubbing genuinely changes comprehension |
| Regulated category (finance, health, legal) | Local script, local review, then record | The regulatory review costs more than the production either way |
Where AI dubbing predictably fails
Failures are not random. Across dubbed output the same categories break, which makes them a checklist rather than a surprise.
Names. Brand names, product names and people's names get translated when they should be preserved, or pronounced as ordinary words. This is the single most common embarrassment and it happens in the first three seconds.
Numbers and currency. Prices, percentages and dates read out in the wrong convention. "Free for 30 days" and a price spoken in the source currency undermine the offer more than a slightly stiff voice ever would.
Locale variants. The list covers languages and locales, which matters: Mexican and Castilian Spanish, Brazilian and European Portuguese are different products for an audience. Choosing the wrong one is not a translation error, it is a positioning error.
Humour, idiom and wordplay. These translate literally and land as nonsense. If your script's best line is a pun, it will not survive.
Legal and disclaimer language. A dubbed disclaimer may be technically accurate and still fail to meet a local requirement for specific wording.
Pacing against the cut. Languages expand and contract. A German dub of a tight English script often runs long, so lines drift off the shot they were written for.
The organic counterpart, and why the label matters
If you also run a YouTube channel, you have a parallel decision with different rules, and they are easy to confuse.
| Google Ads dubbing | YouTube auto-dubbing | Custom multi-language audio | |
|---|---|---|---|
| Applies to | Paid video creative in Asset Studio | Organic uploads | Organic uploads |
| Coverage | 33 languages and locales | 27 languages, expressive speech in 8 | Whatever you produce |
| On-screen label | Standard ad disclosure applies | Marked "auto-dubbed" to viewers | No auto-dub label |
| Quality control | Review before launch | Limited | Full |
| Best for | Scaling performance video into new markets | Reach on long-tail library content | Brand channels and flagship films |
The "auto-dubbed" label is the detail brands underestimate. On a piece of library content it is a fair trade for reach. On your flagship brand film it tells every viewer in that market that they are watching the version you did not care enough to make, which is the opposite of what a brand film is for. Uploading your own audio tracks avoids it. That trade-off sits alongside the wider question of what brand video actually costs and what drives the number.
Disclosure, consent and the part that is not about performance
Platform labelling of AI-generated ad assets and regional transparency rules have both tightened through 2026, and a synthetic voice delivering a brand message is squarely the kind of asset they address. Check the requirement in each market you are entering rather than assuming the platform handles it, and keep a record of which assets were AI-generated. We covered the labelling landscape in more detail in our piece on Google and Meta's AI ad disclosure labels.
The sharper risk is consent rather than compliance. Dubbing a testimonial, a customer interview or a named employee generates a recording of that person saying something they never said, in a language they may not speak. Even where it is lawful, it is the kind of thing that reads badly when it surfaces. Get written permission that covers synthetic voice specifically, or subtitle those pieces instead.
A nine-point check before a dubbed ad goes live
This takes about twenty minutes per language and it is the difference between the feature saving money and costing reputation.
- A native speaker of the target locale has watched the whole ad, not read the script.
- Brand and product names are pronounced correctly and are untranslated.
- All prices, dates, percentages and units follow local convention.
- On-screen text, supers and the end card are in the same language as the audio.
- The call to action matches what the landing page actually says.
- The landing page, form and confirmation are in that language.
- Someone can answer an enquiry in that language within your normal response time.
- Legal and claim language has been reviewed for that market.
- Disclosure requirements for AI-generated assets have been checked, and consent exists for any identifiable person.
Items 6 and 7 fail most often, and neither is a creative problem. They are the reason we treat market entry as a funnel decision rather than a creative one — the same logic behind building landing pages specifically for paid traffic.
How to test it this quarter
Pick one market where you already have some demand signal and can support the language. Localise the landing page first. Take one voiceover-led creative that is already working in English, dub it, run the nine-point check, and launch it against the English version in the same market with the same budget and bid strategy. Give it enough time to clear the learning period and judge it on cost per qualified outcome rather than click-through rate, because click-through rate is precisely the metric a language match flatters most.
If it wins, expand one language at a time and re-run the check each time. If it does not, you have learned something more valuable than a dubbing decision: that the market's constraint was never the language of the ad.
One structural change is worth making regardless of the test result. Scripts and edits can be built to localise: keep on-screen text on its own layer rather than baked into the footage, avoid puns and idiom in the lines that carry the offer, leave a little air in the pacing so a longer language still fits the cut, and keep prices out of the voiceover where you can. None of that costs extra at the point of production, and it is how we scope ad films and brand video for clients who expect to run in more than one market.
Frequently asked questions
What is Google Ads video dubbing in Asset Studio?
It is an AI feature inside Google Ads Asset Studio that translates and dubs an existing video creative into other languages, covering 33 languages and locales. It produces a new audio track for a video you already uploaded, so one master creative can run in multiple language markets without a separate production. It is rolling out to selected advertisers and is free for those who have it.
Is Google Ads AI dubbing free?
The dubbing itself is free for the advertisers who currently have access, but localising a campaign is not. The costs that remain are legal and claims review in each market, re-rendering on-screen text and price supers, a native-speaker check of the output, and localised landing pages and offers. In our experience those four items, not the voice track, are where most of a localisation budget goes.
Should I use AI dubbing or subtitles for video ads?
Use dubbing when the message lives in the voiceover and the viewer's attention is on the visuals, which is typical of product demos and founder-to-camera ads on connected TV and YouTube. Use subtitles when the ad plays muted by default, when the speaker is on camera in close-up so lip mismatch is obvious, or when exact wording carries legal weight. Many advertisers should do both: a dubbed track plus localised subtitles.
How accurate is Google's claim that dubbing lifts click-through rate by 86%?
Treat it as a vendor case-study figure rather than a benchmark. The 86% click-through-rate increase and 75% cost-per-click reduction come from Google's own promotional material for the feature, without a published methodology, sample size or baseline. The direction is plausible, because an ad in the viewer's language should outperform one that is not, but the magnitude should not be used in a business case. Measure it in your own account instead.
What is the difference between Google Ads dubbing and YouTube auto-dubbing?
Google Ads dubbing applies to paid video creative inside Asset Studio, covering 33 languages and locales. YouTube auto-dubbing applies to organic uploads, is available in 27 languages, adds an on-screen "auto-dubbed" label and, since early 2026, expressive speech in eight languages. For organic brand channels, uploading your own multi-language audio tracks avoids the auto-dubbed label and keeps quality under your control.
Do dubbed video ads need an AI disclosure?
Assume yes and check your markets before launching. Platform AI-labelling rules and regional transparency regulation have been tightening through 2026, and a synthetic voice speaking on behalf of your brand is exactly the kind of asset those rules target. The practical risk is not usually a platform penalty but an endorsement problem: a dubbed testimonial puts words the speaker never said into their voice, which is a consent question before it is a compliance one.
The takeaway
Asset Studio's dubbing removes a real barrier: the cost of recording a voiceover was the reason most advertisers never tested a second language, and that reason is gone for the creatives it suits. Use it on voiceover-led performance video where the visuals carry no language, keep it away from close-up talking heads and testimonials, and never let a dubbed ad point at a landing page in a different language than the one it is spoken in. The thing worth remembering when the 86% figure gets quoted in a meeting is that it is Google's number for Google's product. Run one honest paired test in one market you can actually serve, and you will have a number that belongs to your business instead.