A question the AI era has made impossible to ignore
Twenty years behind the microphone as an independent French voice over artist, and one foundation that never moved: trust. Trust in the human relationship, in the framework agreed with the client, in the use that will be made of every delivered recording. For a long time, that model was enough.
Today, it is cracking. Content travels faster than contracts, distribution channels keep multiplying, and one question is rising across the whole profession:
What really happens to a production featuring a voice over, once it has left the studio?
Why this question can no longer wait
Historically, usage was readable: a voice recorded for a TV campaign, a corporate video or a radio spot, with an exploitation scope defined in black and white. The file went out, lived the life set out in the contract, end of story.
That time is over. Content gets repurposed, reshared, rebroadcast, sometimes transformed, rarely out of malice, but with a loss of visibility at every step. And AI has changed the scale of the problem: a few seconds of audio are now enough to reproduce a timbre, an intention, a vocal identity.
The issue is not the ethics of clients, real and sincere in the vast majority of cases. The issue lies elsewhere: we no longer know how to follow the life of a voice over time.
When a production raises questions, after the fact
Some time ago, I came across a TikTok video. A cable car near Paris, passengers filming, and in the middle of it all, an audio safety announcement. My voice. The timbre recognised within the first second, no hesitation possible.
Impossible, on the other hand, to remember when I had recorded that message, for which client, under which contract. No dispute on my side, no blame on my partners, just a factual observation: no visibility left on the life of that production.
And there is nothing exceptional about this situation. Once delivered, an audio production travels far beyond what any of us can track individually.
A few months later, the story went much further than that grey area. In spring 2026, I found my voice cloned without my consent on Fish Audio, an AI platform, on two pirate accounts, one of them named after me. Self-service, ready to say anything to anyone. And not only mine: my daughter Florine’s voice was running there too, cloned under a pseudonym, with nearly 14,000 uses. The full story, and the exact procedure to get a clone removed, are in My Voice Was Cloned and Stolen on Fish Audio.
And the most revealing part of this affair: these platforms verify no voice print, ask for no proof of consent before cloning a voice. Anyone can upload anyone’s voice, the door is wide open. Yet they know perfectly well how to read an audio fingerprint, it is exactly what they do to check a claim after the fact. The verification exists, it is simply placed on the wrong side: after the theft, never before.
A collective issue, not an isolated case
Tamper-proofing and traceability do not only concern high-profile or heavily exposed voices, they run through the entire market. The ease with which a voice can be copied, transformed or rebroadcast forces us to rethink our relationship to audio files, to usage rights and to vocal identity itself.
Nothing anxiety-inducing here. It is a debate about professional maturity.
Tamper-proofing and traceability
What are we actually talking about?
Two notions that sound technical, and are actually very simple.
Tamper-proofing is the guarantee that an audio file cannot be modified, altered or reused without it being detectable. Traceability is the ability to know where a production is broadcast, when, and how it is used, throughout its life.
Several approaches already exist:
- Audio watermarking: an inaudible marker embedded in the file, which makes its use trackable without changing anything to the listening experience.
- The audio fingerprint: a unique signature, specific to a voice or a production, which identifies it reliably.
- Enriched metadata: information attached to the file, travelling with it from broadcast to broadcast.
- Blockchain-inspired technologies: designed to guarantee the integrity of a piece of content and keep the history of its uses.
These notions, and all the vocabulary that goes with them, are detailed in the voice over glossary. To date, none of them has emerged as a universal standard. And that is precisely where the challenge lies.
What is still missing
Three things are missing today: a shared vision between voice talents, studios, agencies and advertisers, simple tools that respect creation and the human relationship, and a framework that protects without surveilling.
Traceability must never become a tool of control. Well designed, it is the opposite: a lever for understanding campaigns, respecting usage rights and securing practices, for everyone.
A conversation already under way in the profession
These questions are not isolated. In France, initiatives such as “Touche pas à ma VF” (“Hands off my French dub”) have brought voice protection issues into the public arena. Internationally, the National Association of Voice Actors has been working on these subjects for years, AI first.
And on the regulatory side, the discussions around the AI Act show that all of this goes far beyond the artistic sphere: it is law, technology and ethics at the same time.
Opening the dialogue: the spirit of the Apéro Voice
This is why I dedicated an Apéro Voice to these questions on 28 January 2026, with Loïc Thaler, a Flemish voice over colleague who has spent the past two years developing an application dedicated to voice traceability. The goal was not to sell a ready-made solution: informing, raising awareness, bringing people together, sharing a state of play, and thinking together about a more readable future for our profession.
The Apéros Voice are free, held in French, and open to everyone who makes a living with their voice. To discover Loïc’s work: Loïc Thaler, Flemish voice over artist.
Why talk about it now
Because everything is moving very fast, because usage evolves faster than the existing frameworks, and because silence feeds confusion. These subjects deserve time, nuance, dialogue, and open, constructive professional spaces to discuss them.
What now?
If you are a voice talent, a studio, a producer or an advertiser, this conversation already concerns you, even if it has not yet taken shape on your side. Subscribe to the newsletter to follow these discussions, or join the next Apéro Voice to talk about it in person.
The traceability of audio productions is not a constraint. It may well be one of the pillars of a profession that is more readable, better protected and better respected.
Estelle Hubert French voice over artist