Do you need consent to clone a voice for dubbing?

If the voice is yours, no. If it belongs to someone else, yes, and increasingly that's a legal requirement rather than an ethical preference.

But the question hides a category error to clear up first, because "cloning a voice" and "dubbing a video" have been merged by the market and they aren't the same operation. Some tools ask you to enrol a voice: record samples, build a reusable profile, then generate any speech you like in it. Dubbing takes an existing recording and produces a translated version of that specific recording.

The first creates a reusable instrument out of a person's voice. The second translates a thing they already said. The consent analysis is different, and so is the risk.

Three separate questions, and people routinely answer one and assume they've covered all three:

• Consent attaches to the voice. Whose voice is being reproduced, and did they agree?
• Rights attach to the content. Do you hold the copyright in the material you're dubbing?
• Disclosure attaches to the publication. Does the platform or the jurisdiction require you to say it's synthetic?

This is not legal advice. Anything involving another person's voice commercially is worth a lawyer's half hour.

What the law now says about voices

The legal position has moved fast, and voice is no longer an afterthought to image rights.

Tennessee's ELVIS Act (Ensuring Likeness Voice and Image Security) was signed on 21 March 2024 and took effect on 1 July 2024. It was the first US state law written specifically for AI voice replication, and it made a person's voice a protected property right in its own terms. Critically, it covers not just an actual voice but a simulation of one, and violations carry both civil liability and criminal exposure as a Class A misdemeanour.

That "simulation" wording is the part that matters technically. A defence along the lines of "it isn't really their voice, a model generated something similar" is exactly what the statute was drafted to close off.

Other states have followed with their own right-of-publicity updates, and a federal bill, the NO FAKES Act, has been under discussion to make the standard consistent nationally rather than state by state. Treat the patchwork as the current reality: which state's law applies can depend on where the person lives, not where you are.

In the EU, the framing is different. The AI Act's Article 50 obligations, applicable from 2 August 2026, deal with transparency rather than ownership: a deployer generating a deep fake has to disclose it, where a deep fake means content resembling a real person that would falsely appear authentic. That's a disclosure duty, not a consent duty, and it sits alongside rather than instead of existing personality and data protection rights.

What this means for dubbing your own content

Almost none of it applies, and that's not a technicality.

If you're dubbing yourself, you're the person whose voice is involved and you've consented by uploading it. The right-of-publicity statutes protect you against others, not against yourself. The EU disclosure obligation turns on content that "would falsely appear to a person to be authentic", and a translation of your own words genuinely isn't that. YouTube's altered content policy explicitly exempts cloning your own voice for voice overs and dubs.

Our product also has no voice-enrolment step at all, which changes the analysis in a way worth understanding. There's no "record three sentences" screen, no stored voice profile, and nothing that could be reused to make you say something you never said. The upload takes a file and one of 32 target languages, and the submission carries the file, the target language and one flag.

So there's no persistent artefact of your voice to consent to the ongoing use of. There's a file you uploaded, which is deleted as soon as the dub is produced and within a day regardless, and an output you keep.

When you do need consent

The cases are more common than people assume, because "my content" and "my voice" aren't the same set.

Anyone else speaking in your video. An interview subject, a guest, a colleague in a training video, a customer in a testimonial. You may hold copyright in the recording while they retain rights in their voice and likeness. Dubbing produces them speaking a language they may not know, saying words they never said in it.

Employees and contractors. A release covering use of a recording is not automatically a release covering synthetic reproduction of a voice in other languages. Older agreements simply won't have contemplated it.

Anyone recognisable at all. The right-of-publicity risk scales with identifiability. A narrator nobody could pick out is a different exposure from someone with a distinctive, commercially valuable voice.

Public figures, always. Both the legal risk and the disclosure obligations are at their highest, and the artistic and satirical carve-outs in the EU rules are narrower than people hope. "Evidently satirical" means evidently.

Getting consent isn't complicated in most of these cases. Ask, in writing, and be specific that it covers AI-generated translated versions rather than just the original recording.

When consent isn't the issue but rights are

Consent covers the voice. It does nothing about the content.

Our terms require that you don't upload copyrighted content without authorization, in those words. Permission from a speaker to reproduce their voice is not permission from a rights holder to make a derivative of their work, and a dub is a derivative work. Both can be required at once, from different people.

We've covered that side in who owns an AI-dubbed track, including the cases people get wrong: commissioned video, licensed music, and footage used under fair use.

A practical order of operations

Before dubbing anything with another person in it:

Establish who is audible in the file, and whether each of them has agreed to synthetic translated versions of their voice specifically. Establish who owns the underlying content, which is often somebody else again. Then check what the destination platform requires you to disclose, which is a separate question with a separate answer, covered in when you have to label an AI-dubbed video.

If any of the three comes back unclear, the cost of resolving it now is a conversation. The cost of resolving it after publication, in a jurisdiction that treats voice as a property right with criminal penalties attached, is considerably higher.

For training and course content, where multiple speakers and older contracts collide most often, dubbing e-learning from English to Italian covers the practical workflow, dubbing corporate video from English to Portuguese covers the internal-comms equivalent, and the use case index has the rest.

FAQ

Do I need consent to clone my own voice?

No. You're the person the protections exist for, and uploading your own recording is your consent. Dubbing your own content also falls inside YouTube's disclosure exemption and outside the EU's deep fake definition.

Can I dub someone else's video?

Two permissions are needed and they come from different people: the rights holder in the content, because a dub is a derivative work, and the person whose voice is reproduced. Holding one doesn't give you the other.

What is the ELVIS Act?

Tennessee legislation, effective 1 July 2024, that made a person's voice a protected property right and covered simulations of a voice rather than only actual recordings. It carries civil liability and criminal exposure, and it set the template other states have followed.

Does an old talent release cover AI dubbing?

Often not. Releases written before synthetic voice reproduction existed generally cover use of the recording rather than generation of new speech in that voice. If it matters commercially, get a fresh release that names AI-generated translated versions explicitly.