OpenPhn docs
Concepts

Voice cloning

Use a custom cloned voice — gated behind a recorded biometric-consent attestation.

OpenPhn can place calls in a custom cloned voice instead of one of the stock voices. Because a voice is biometric data, a clone is not usable until you record an attestation of consent from the person whose voice it is — there is no path to provision a clone without it.

Lifecycle

  1. Create the clone (name + the source voice sample, provisioned via Cartesia).
  2. Attest consent — POST /v1/voice_clones/{id}/attest. The clone is provisioned with the voice provider as part of this step, so until it's attested it cannot be selected anywhere.
  3. Use it — set the clone as a number's voice (voice_clone_id on the number) so calls on that line speak in the cloned voice.

The attestation

The attestation records who consented and where the proof lives, and is retrievable as a signed record for your audit trail:

POST /v1/voice_clones/{id}/attest
{
  "attestor_name":     "Jane Doe",
  "attestor_email":    "jane@example.com",
  "attestation_text":  "I, Jane Doe, consent to OpenPhn cloning my voice…",
  "consent_form_url":  "https://docs.example.com/signed-consent.pdf"
}
  • GET /v1/voice_clones/{id}/attestation returns the signed attestation (scope numbers:read). Recording or modifying attestations requires numbers:write.

Cloning a person's voice without their documented consent may violate state biometric-privacy laws (e.g. Illinois BIPA) and platform policy. The attestation gate exists to make consent a hard prerequisite, not an afterthought — keep the signed consent form referenced by consent_form_url on file.

On this page