In short: We have cloned one voice in public. The Phoenix Formula, our audio course on creativity, is partly voiced with a consented cloned voice and labelled as synthetic. UK law has not given performers a statutory right over their voice, so consent, scope and labelling all sit in the agreement.
The Phoenix Formula is the one place we have used a cloned voice in public, and it is a small case on purpose. The course is presented by Mia Moore and partly voiced with a consented cloned voice, and the course says so. This page sets out what that case looks like, what the law leaves to the contract, and where our own published rules stop.
The case in brief
- The product: an audio course on creativity with a short welcome and 24 weekly episodes, each with a short PDF workbook, delivered as a private podcast feed. It is priced at £149 including VAT.
- The source: the course was built from the archive of Creative Conversations, the podcast hosted by Sophia Gray and Mia Moore.
- The voice: the course is partly voiced with a consented cloned voice.
- The label: the cloned voice is labelled as synthetic.
That is the whole of what we publish about it. We are not describing the production method here, and we are not offering the case as proof that every clone job goes well.
Why we started from the paperwork
On 28 August 2026 more than 80 performers wrote to the Prime Minister asking for a statutory right over their own voice. A government spokesperson replied that it would consult on how to address the harms while protecting legitimate innovation. As at 28 September 2026, the consultation on digital replicas had not opened.
So the law has not moved, and the protection a performer actually has is whatever is written into the booking. That was the starting point for the consent. The agreement had to do the work that a statute might do later.
The five questions, applied to a clone
Our write-up for performers lists five questions to ask any studio before a session. For a cloning job, each one changes slightly.
- Who keeps the recordings, and for how long? The raw session, every take and the edited masters. Name each one and give a date.
- What is the recording licensed for? Which media, where and for how long. A voice model needs its own use named, not just the recordings it was built from.
- Can it be used to build a synthetic voice? For an ordinary booking, ask for a plain no in writing. For a cloning job the answer becomes a yes, in writing, for the named use, and with the performer agreeing to it and being paid for exactly that.
- What happens at the end? Deletion, when it happens, and who confirms it. This matters more for a model than for a file, because a model can be reused without anyone re-recording.
- How do the files travel? A private link that expires leaves less lying about than a shared folder nobody remembers setting up.
None of this needs a lawyer to ask. It needs asking before the session, while everyone is still in a good mood.
What gets labelled, and when
Labelling is the part the listener sees. The Phoenix Formula labels its cloned voice as synthetic. The same rule applies on the performer's side: our write-up for performers says that if a studio plans any AI process on a recording afterwards, it should say so before the session, not after.
A read recorded under clear terms is a read nobody argues about later. That is as true for the producer who commissions it as for the performer who gives it.
What we do not claim
We publish the rules for our ADR and dialogue room because unreleased picture comes through it. We sign your NDA before any picture is sent. Phones stay out of the room while unreleased picture is on screen. Picture and takes are kept for 30 days, then deleted. Files go by a private link that expires.
Those rules were written for picture security. They are not an AI policy, and we will not dress them up as one. They cover the first question on the list and most of the last two, which is the level of detail to ask any studio for, in writing, before a session.
If you are thinking about your own voice
Whoever you talk to, ask the five questions first and read the voice and likeness consent checklist. For the wider picture, see who is responsible for an AI-made ad and our earlier note on what performers and producers can do now.
Sources
- Granary Digital, Voice cloning has reached Downing Street. What performers, and the people who hire them, can do now, read 5 November 2026
Related guides
- AI video for small businesses: what's real, what's risky
- When to use AI and when to shoot it for real
- AI in the production workflow, stage by stage
- Who is responsible for an AI-made ad?
- Voice and likeness: a consent and contract checklist
- All guides and tools
The picture at the top of this page is an illustration, not a photograph of a client or a real job.
Questions we get asked
Can a studio clone my voice without asking?
It should not, and, as at September 2026, no UK statute gives performers a specific right over their own voice. The protection is the paperwork: a written agreement saying what the voice model is for, where it is used, who keeps the recordings and when everything is deleted.
Does a cloned voice have to be labelled?
On our own course, it is labelled as synthetic, and we think that is the right standard for any listener-facing use. Tell people before they press play, and tell the performer before the session that a synthetic voice is the plan.
What should a voice cloning agreement cover?
Who keeps the recordings and for how long, what the recording is licensed for, whether it can be used to build a synthetic voice, what happens at the end of the agreement, and how the files travel. These are the five questions in the list below.
granary.digital · 01243 884993 · studio@granary.digital · The Old Grainstore, Adsdean, Chichester, West Sussex, PO18 9DN · Prices exclude VAT unless marked inc VAT