When a voice clone is
the right choice.
On the memorial concierge tier, what voice work involves, the consent framework, and the conversations we have with families before we agree to begin.
The Memorial Concierge tier of our studio offers something we don't recommend to most families: a 90-second letter, written by the family, read in the voice of someone who has died.
This is the only piece of work in the studio where we decline more orders than we accept. A short, honest note on what it is and what we won't do.
What it is.
A professional voice-cloning service can build a custom voice model from roughly 30 seconds of clean source recording — a voicemail, a recorded prayer, a wedding-video clip. The model can then read text aloud in that voice.
The text we read is a letter your family writes. It's typically addressed to one or more living relatives. It's read once, played at the memorial service, and sometimes shared privately afterwards with family members who couldn't attend.
When it lands well.
The version of this work that lands well, every time we've done it, has three properties:
- The letter is something the person could plausibly have written. Not "what they would have said about [current event]" — what they did say, in the register they used, addressed to people they knew.
- The whole family is on side. If one sibling is uncertain, we wait until they're not.
- The audience is small. The immediate family, at the memorial service, with a private link for relatives who couldn't be there. Not "we want to share it on social media" — that's a different conversation.
When we say no.
Common reasons we'll decline, with a refund and a written explanation:
- The deceased was on record opposing AI use of their likeness.
- A close family member is opposed or unaware of the work.
- The intended text fabricates events, opinions, or words the person didn't say or hold.
- The intended audience is broader than the immediate family — broadcast, publication, public memorial pages, etc.
- The legal review for your specific case takes longer than the date you need it by.
The UK legal picture.
UK law on this is cleaner than the equivalent US framework. The UK GDPR doesn't apply to deceased persons (Article 27 / ICO guidance). The UK has no statutory right of publicity. The legal levers are copyright in the source recording (we need to have the right to use it), defamation (the voice can't be made to say something defamatory), and contract (we work from a signed Voice Clone Use Agreement with the commissioning relative).
We have all our voice-clone work reviewed by a UK solicitor before delivery. The framework documents are public — see our privacy notice for the data-protection side and the Voice Clone Use Agreement template at intake.
What it costs and what arrives.
The Memorial Concierge tier (£1,249) includes 20 photographs restored, two framed pieces (16×20 + 11×14, walnut), an extended slideshow video, the voice-clone letter, a presentation box, a handwritten delivery note, and a studio direct line during the planning week.
The voice-clone work specifically is delivered as a 90-second audio file. We don't retain the trained voice model after delivery — it's deleted within 30 days. We don't reuse the model for any other order or purpose.
To discuss a memorial concierge order, write to hello@keptandkin.com with "Concierge" in the subject. We answer the same day and we'll be honest from the first email about whether this is the right choice for your family.