AI Voice Cloning for Dubbing: What You Can and Can’t Do

You can clone a voice for dubbing when it is your own voice, or when the person has clearly agreed to have their voice cloned for that purpose. You cannot clone someone just because their speech is in your video: a guest, an interviewee, a celebrity in a clip, or a colleague who left the company. Consent is the line almost every law and every dubbing provider draws. Get it in writing, keep it specific, and tell viewers when a voice is synthetic.

This article is general information, not legal advice. Rules differ by country and change quickly; for commercial projects with other people’s voices, ask a lawyer in your market.

The short version

When the answer is no, you can still dub: give that speaker a natural voice from the library instead. Viewers get the content in their language without anyone’s voice being copied.

Why voice cloning needs consent

A voice identifies a person as clearly as a face. A clone can say things the person never said, in a voice their audience trusts. That is why the law treats voice data seriously and why every reputable provider requires permission.

In the EU: your voice is personal data, sometimes biometric

Under the GDPR, a recording of someone’s voice is personal data. Voice data used to identify a person counts as biometric data under Article 9, which is prohibited unless an exception applies, most commonly explicit consent for a specified purpose. In practice, a cloning consent should name the purpose (for example, “dubbing our course videos into Spanish and German”) and should be as easy to withdraw as it was to give.

The EU AI Act: tell people the voice is synthetic

Article 50 of the EU AI Act sets transparency rules for synthetic content. Providers of AI systems must mark generated audio in a machine-readable way, and anyone who publishes a “deep fake”, content that resembles real people and would falsely appear authentic, must disclose that it was artificially generated. For clearly artistic, satirical or fictional work, the disclosure can be light, but it is still required. The rules apply from August 2026; check the current timetable, as parts of the AI Act have been subject to amendments.

In the US: a patchwork that is getting stricter

• Illinois (BIPA): voiceprints are biometric identifiers. Collecting them needs a written release, and violations can cost $1,000 to $5,000 each.

• Tennessee (ELVIS Act, in force since July 2024): adds a person’s voice, including a simulated one, to their right of publicity, and can reach those who make tools for unauthorised replicas.

• The FCC ruled in February 2024 that AI-generated voices count as “artificial” voices under the TCPA, so robocalls using them need consent.

• Federal NO FAKES Act: reintroduced in 2026 to protect voice and likeness nationally, but not law as of October 2026.

Elsewhere

Denmark has introduced a bill that would give people rights over realistic digital imitations of their face and voice, planned to take effect in 2027. Expect more countries to follow. Consent-first habits will keep you on the right side of new rules as they arrive.

Providers set their own rules too

Voice providers ban cloning without permission in their terms. ElevenLabs, which generates the voices for BlaBlaScribe dubbing, prohibits cloning or impersonating the voice of any living person without their explicit and informed consent, and bans voice recordings of minors, according to its prohibited use policy. Breaking those terms can cost you the account, whatever the local law says.

A practical consent checklist

• Ask before you record, or at least before you dub. “Can we clone your voice to dub this episode into Spanish?” is a short question.

• Get it in writing: who, which recordings, which languages and channels, and for how long. An email reply is better than a verbal yes.

• Keep it specific. Consent to dub one series is not consent to create new lines for ads.

• Make withdrawal easy, and delete the clone when someone withdraws.

• Don’t change the meaning. A dub translates what the person said. Rewriting their words into something new is a different use that needs its own permission.

• Disclose. A line in the description such as “Dubbed with AI; voices are synthetic” covers viewers and platform rules.

• Keep a record of the consent next to the project files.

How cloning works in BlaBlaScribe

In BlaBlaScribe AI Dubbing, a clone is made only from one speaker’s speech in the file you are dubbing, and it needs at least 20 seconds of that speech. Before each clone you confirm: “This is my voice, or I have the speaker’s permission to clone it and use the copy for voice-over.” The clone costs the same as any library voice, is stored in your account for reuse, and can be deleted at any time. Clones unused for 12 months are deleted automatically. Voices are generated by ElevenLabs; transcription runs on our own servers in Germany. Details are in the privacy policy.

If you can’t get permission, pick a library voice for that speaker. You can filter by female or male voices and listen to samples, and each speaker keeps a consistent voice through the whole video.

When a library voice is the better choice anyway

• Interviews with many guests: collecting consent from everyone takes longer than choosing voices.

• Short or noisy speech: a clone needs at least 20 seconds of clear speech from that speaker.

• Brand voice: a consistent narrator across all your videos can matter more than matching each presenter.

Frequently asked questions

Is it legal to clone my own voice?

Yes. Your own voice is yours to use. Still label the dub as AI-generated where platform rules or the EU AI Act require it.

Can I clone a guest’s voice if they signed a release to appear?

Usually not. A standard appearance release covers recording and publishing, not creating a synthetic copy of the voice. Ask for consent to cloning specifically.

Do I have to tell viewers the voice is AI?

In the EU, synthetic content that could pass as a real person speaking must be disclosed under the AI Act. Platforms such as YouTube also ask creators to label realistic altered or synthetic content. A one-line note in the description is a simple habit.

Can I clone a famous voice for a parody?

Parody may be protected in some countries, but voice providers generally prohibit cloning real people without consent, and right-of-publicity laws can still apply. Use a library voice.

Can I dub without cloning at all?

Yes. Choose a natural voice for each speaker from the library. Our step-by-step dubbing guide covers both options, and the video translator and subtitle translator cover translation with subtitles only, with no voice involved.

The bottom line

Clone your own voice freely. Clone anyone else’s only with specific, written consent. Disclose synthetic voices to viewers, and use a library voice whenever consent isn’t possible. Dubbing doesn’t require cloning; it only requires a voice that suits the speaker.