01 logo

Talking Turkey With Your Voice

Yes, they could create your voice with AI. Deepfaking could go further, maybe even get you into the charts

By James MarineroPublished 3 years ago 3 min read
Spectrogram of the spoken words “nineteenth century”. Frequencies are shown increasing up the vertical axis, and time on the horizontal axis. The legend to the right shows that the color intensity increases with the density.. Image credit: By Aquegg — Own work, Public Domain, https://commons.wikimedia.org/w/index.php?curid=5544473

What happens if a computer system copies your voice? Do you have any legal protection?

It may not be an issue for me, but, if say you’re Tom Jones or Beyonce then you might object to a song being released with an artificially generated voice that sound exactly like you.

‘Stars’ of stage and screen, sports and and other celebrities hold (or try to hold) image rights. But voice rights? Is there such a right?

I know little about the legalities of such things, but a recent announcement by Spotify might be setting the tone for such a scenario.

Audiobooks and more

Spotify acquired Findaway, a ‘leading’ audiobook platform, in November 2021.

And in a recent buy they snapped up Sonantic, a London-based startup which is focused on building AI software capable of generating completely made-up voices.

‘Completely made up’. I bet.

It's expensive to employ a good voice to read a book into an audio file. I know - I've looked at it for my own books. So this is a good way of increasing profits and possibly attracting more authors, like me.

Training

AI systems have to be trained on representative data. Machine learning is the process. So what do they train this Sonantic AI system with? Yes, voices.

Human voices, not dogs or parrots (now there’s a thought).

Whose voices?

Well, I don’t know about you, but there are no recordings of my voice out there (well, maybe one, but that’s another story). So what voices are out there?

I could have picked speeches by other great leaders, such as Trump or Putin, but you get the picture, you'd want to fight them on the beaches, to be sure.

Then there are singers, by the thousand.

And bad singers doing karaoke.

Real or fake?

If a voice is generated by this AI software, how can anyone be sure that it is not already in use by a human being? Think about it.

Plagiarism can be checked online; music even. Photos I guess, to some extent.

Surely it would be a nice, energy and time saving — and profitable — approach just to copy someone’s voice and say it was machine generated?

The legals

I already said that I’m not a legal person, but I’m going to have a barrack room lawyer’s stab at this case.

Say that the Litigant in Chief, Donald Trump, has his voice generated by a machine — a pure coincidence of course — and the voice is used to read a speech.

Where would that leave Trump?

Could he sue?

And, if he did, how would he prove that the machine voice was a copy of his?

Voiceprints? Maybe he could show identical voiceprints. But would he have rights over his artificially created (not copied) voice? I guess that there’s an argument along the lines of

‘Well, your Honour, the plaintiff never spoke those words, so however it sounds, voiceprint or not, it is not the plaintiff speaking. The plaintiff is not passing off as Donald Trump, in fact, quite the opposite. We all know that Mr Trump is in favor of widespread ownership of automatic firearms and would never criticise the NRA. Nothing has been copied. There is no case to answer, this is a vexatious litigant.

Precedent

There’s precious little precedent in this area, as it’s new. AI is still learning.

There was a case reported in 2021 where TikTok was being sued for breach of voice rights, but I have been unable to discover whether it has been resolved.

From what I have gleaned, a person’s voice rights — even in the context I am considering here— can be protected in a contract with an organisation for which that person is working in some capacity.

So, a newsreader working for CNN could protect his/her voice rights in his/her contract with CNN, even to the extent that CNN could not use an AI voice generator to ‘put words into that newsreader’s mouth’.

But that requires a contract.

Music

In the case of music, Ed Shearan recently won a copyright case when he proved (as I read it) that ‘Shape of You’ did not contain plagiarised content, that it was his creation and that any similarity to the work of other artists was coincidental.

Deepfaking

It seems to me that provided an AI generated voice which closely resembled that of a living person did not purport to be of that person, then it would be open season. Don’t fake it, just use it.

Disclaimer: I’m far from being a legal eagle, so please don’t rely on the content of this article for making important career or artistic decisions. Or speeches. If you do, the outcome could be messy.

Tom Jones and The Cardigans ‘Burning Down The House’. For real.

***

No audiobook versions yet!

James Marinero on Gumroad

***

Canonical: This story was first published in Medium on 23 June 2022

futuregadgetsproduct reviewstartuptech newsvr

About the Creator

James Marinero

I live on a boat and write as I sail slowly around the world. Follow me for a varied story diet: true stories, humor, tech, AI, travel, geopolitics and more. I also write techno thrillers, with six to my name. More of my stories on Medium

Reader insights

Be the first to share your insights about this piece.

How does it work?

Add your insights

Comments

There are no comments for this story

Be the first to respond and start the conversation.

Sign in to comment

    Find us on social media

    Miscellaneous links

    • Explore
    • Contact
    • Privacy Policy
    • Terms of Use
    • Support

    © 2026 Creatd, Inc. All Rights Reserved.