Voice and recording
Cloning a voice with permission: a checklist before you begin
What to agree on regarding purpose, the person whose voice is used, and how the audio is presented before using someone else's sample.
Having access to a recording does not mean you have permission to clone the voice it contains. The technology can generate sentences the person never spoke, so consent and how the result is presented are important. Before using someone else's sample in Vox, agree with the person whose voice it is on what it will be used for and what limits will apply to that use.
Clarify the purpose
Explain that Tanakh narration will be generated and that the result will be synthetic. Do not describe it as a recording of the person reading each passage. If the intended use changes, consult them again rather than assuming the initial permission covers any purpose.
It is advisable to keep a clear agreement about the use, especially if the audio will be shared beyond a personal setting. This guide offers practical precautions, not legal advice or a guarantee of compliance in every country.
Check whose voices appear in the sample
Use only one voice. If other people can be heard, avoid using that recording and prepare a different sample. Do not clone the voices of public figures or anyone who has not authorized the purpose simply because you find a file online.
For the voices of deceased people, rights and family agreements may need to be reviewed. Do not assume you have authorization because you are related or have kept an old recording.
Present the result without misleading anyone
Make it clear that the audio uses a cloned voice and does not establish that the person whose voice it is has read or approved every sentence. Do not use it to simulate statements, consent, or endorsement by that person either.
Cartesia documents the technical capability for cloning, but that capability does not settle questions of permission. In Vox, prepare the 10 seconds carefully and test an excerpt. Personalization is more responsible when it combines a clear sample, an agreed purpose, and an honest explanation of the audio's origin.