Yes — with roughly 60 seconds of usable recordings. Modern voice models can learn a person's voice from home videos, voicemails, and voice messages, then read new words in that voice: a farewell letter, a blessing, a thank-you. The result is not a recording that exists somewhere; it is a reconstruction — which is exactly why the family, not the studio, should be the judge of whether it truly sounds like them.
At Voice Legacy, that judgment is built into the process: we produce a 15-second preview first, and if the family says it doesn't sound like their person, the project ends and nothing is charged. In our experience, that single rule answers most of the anxiety people bring to this question.
About 60 seconds of them simply talking — and voicemails count. The recordings don't need to be professional. What matters more than length is clarity: one clear 40-second birthday video usually beats five minutes of noisy restaurant footage.
If you're not sure what you have is enough, that is precisely what the preview step is for. We test first; we only take the project if the voice holds up.
Only under conditions — and consent is the first one. Researchers at the University of Cambridge's Leverhulme Centre for the Future of Intelligence warned in 2024 that AI recreations of the dead need consent protocols and design safeguards to avoid real psychological harm. We agree, and we built our studio rules around that warning.
Our full position lives in our Consent & Ethics policy, but the short version: the person's own wishes come first (if they ever said no, we refuse the project); the close family must agree; the use is a single, bounded farewell — not a chatbot, not an avatar, not "digital immortality"; and the recordings and voice model are deleted within 30 days of delivery.
Source: University of Cambridge, LCFI (2024).
At Voice Legacy: from US$649 for a Voice Farewell — a 2–3 minute memorial video where your loved one's recreated voice reads a farewell letter, paired with restored, gently animated photographs. That includes the preview test, one round of revisions, and delivery typically within 72 hours of receiving materials.
For context across the wider market: DIY photo-animation apps run $5–60 but include no voice and no editing; traditional professional tribute videos run roughly $300–1,800 but — as far as we know — no traditional studio offers voice recreation at all. We publish our prices because families planning a service in four days should not have to book a sales call to learn a number.
It cannot rescue audio that isn't there. If the only recordings are a few seconds long, heavily distorted, or drowned in background noise, the recreation may not reach "that's really them" — and we will tell you that before you pay, not after.
It also cannot — and in our studio, will not — hold a conversation. A recreated voice reading words the family chose is a farewell. Software that pretends to be an ongoing version of the person is something else, and we don't build it. A goodbye should be allowed to end.
Planning a service and wondering if the recordings you have are enough? Send us what you have. We'll reply within 12 hours with an honest assessment — and you pay nothing until you've heard the preview and said "that's them."
Start an inquiry