How much does AI dubbing cost per minute, and when is studio dubbing worth it?

How much does AI dubbing cost compared with studio dubbing?

Published prices run from a few dollars to a few hundred per finished minute, because dubbing covers several different jobs: subtitles, an AI voice-over from a translated script, an AI dub that clones the speaker’s voice, and a studio recording with actors. Localisation (localization in the US) is the wider job around them: text on screen, terms, timing and checks.

The ranges disagree because they price different things: the machine pass alone, the machine pass plus a human check, or a cast with direction and mixing. VideoDubbing.com’s 60–90% saving comes from a company that sells the cheaper side: a ceiling, not a forecast.

How do you estimate the price of your own video?

Every tool applies the same arithmetic first. Rask’s FAQ counts one minute of finished translated video as one minute, so a 5-minute video into 2 new languages uses 10 minutes. Then add what is sold separately.

1. Dubbed minutes: video length × number of languages.
2. The rate for your route, from the table above.
3. A review: 10–20% of the project, or $50–$200 an hour, per VideoDubbing.com.
4. Set-up: a glossary at $200–$1,000, metadata translation at $50–$150 a language (VideoDubbing.com).
5. Multipliers: lip-sync, extra speakers, rush delivery.
6. A re-dub when the script changes after the voice exists, which that guide says is charged in full or in part.

What makes AI dubbing and localisation more expensive?

• Languages: each is another full pass, and a regional variant such as Brazilian or European Portuguese can count as a separate one.
• Speakers: voices must be told apart, and people talking over each other are the hardest case.
• Lip-sync: it multiplies minutes at Rask and is the dearest studio tier in VideoDubbing.com’s list.
• Review and terms: a native speaker checks the script, and a glossary keeps product names stable.
• Music and effects: the new voice needs a clean music-and-effects track or a separation step, and 3Play Media advises clean source audio.
• On-screen text: titles, screenshots and lower thirds are edited once per language.
• Rights: consent for each cloned voice, and for human talent the channels and years of use.
• Changes after the voice exists: a rewritten line means a new take.

Language counts in marketing are not language pairs. Rask offers translation in 135 languages but voice cloning in 32. YouTube’s help page lists 20 target languages for an English video, and only English for a French or Russian one. Ask for your exact pair.

Subtitles, AI voice-over or lip-sync dubbing: which route fits which video?

Where the face is the message, lip-sync decides. Where the voice carries it, as in a podcast or webinar, a clean voice track is enough, and VerboLabs puts voice-over 30–50% below dubbing that matches the mouth.

Why does AI dubbing need a native-speaker review?

A machine can produce a fluent sentence that says the wrong thing. 3Play Media, which sells AI dubbing, names publishing without a native speaker’s check as its main practical limitation. YouTube’s help page warns that automatic dubs can mispronounce words and mistranslate proper nouns, idioms and jargon, and that you cannot edit them, only preview, read the transcript, publish, unpublish or delete.

• Brand and product names, and how each is pronounced.
• Numbers, units, currencies and dates.
• Prices, legal wording and any claim that must stay exact.
• Idioms, jokes and the right level of formality.
• Reading speed against text on screen, and the accent for this audience.

Review the translated script before the voice is generated: editing text costs minutes, while a line changed after the audio exists can be charged as a new take. VideoDubbing.com prices a review at 10–20% of the project or $50–$200 an hour.

Can you clone a voice for dubbing, and what do the rules say?

Cloning keeps the same person speaking in every language, and it turns a voice into something you need permission to use. Get a signed release from each speaker naming the languages, channels, period and whether the clone may be reused. Read the tool’s own terms as well: they set limits of their own.

In the EU, Article 50 of the AI Act has applied since 2 August 2026. The Commission says deployers must tell people when they meet a deepfake: AI-generated or altered audio or video that resembles real people and would falsely seem authentic. Fines can reach €15 million or 3% of turnover. Whether a given dub counts, or falls under the standard-editing exception, is a legal question; the Commission’s guidelines carry examples.

What should a dubbing or localisation quote show?

1. The unit: per source minute, finished minute, language or file, and the minimum charge.
2. The route, language by language: subtitles, AI voice-over, cloned voice, lip-sync or actors.
3. The reviewer: native language, which market’s variant, how many passes.
4. Terminology: who builds the glossary and who owns it afterwards.
5. Speakers: how many, and whether voices are cloned, matched or newly cast.
6. Music, effects, on-screen text and subtitle files: what is supplied, separated or replaced.
7. Revisions: rounds included, and the fee for a re-dub after a script change.
8. Rights and labels: consent records, the licence for synthetic voices, how AI dubs are marked.
9. A test: a 2–3 minute clip in your language pair that includes your hardest lines, as 3Play Media recommends.

What would video localisation cost with VITON¹³?

Localisation of an existing video is selected. Add a new master built to localise cleanly, an AI presenter who speaks the new language, extra avatar languages or vertical cuts. The totals add the live starting prices of VITON¹³ packages in both production modes. Studio actors are not part of it.

The localisation package covers the script, the glossary, a caption or voice track and a localised master with a QA record. It takes 3–5 working days.

When is AI dubbing not good enough?

Use actors, or at least a director, when emotion is the product: drama, children’s content, a brand film whose tone is the point. VerboLabs says AI can lack emotional depth and cultural accuracy for story-based content, and 3Play Media lists the same gaps. It is also the wrong tool for overlapping speakers, noisy recordings and claims that must stay exact.

Wait if the source video changes every month; dub the stable version, not the draft. And if you do not yet know that a market wants the content, subtitles are the cheaper question to ask.