Create an Authorized AI Version of Your Voice

Recording-quality requirements

A clean reference recording produces a better, safer voice model. When preparing your sample:

• Record in a quiet room with minimal echo and no background music.
• Use a consistent distance from the microphone and speak naturally.
• Provide a continuous sample that is long enough for the model to learn your delivery.
• Submit audio you recorded yourself or that you are licensed to use.

The consent-verification process

Every clone request completes a documented consent process before the voice is approved:

• The account holder confirms their relationship to the speaker.
• The speaker records a consent statement that includes a unique consent number.
• The statement is transcribed and checked against the submitted profile information.
• Administrators review flagged requests before a voice is activated.

Voice ownership

A cloned voice belongs to the person it was modeled from. The account holder must be that person or must hold documented permission from them. Consent records are retained so ownership can be demonstrated and consent can be revoked.

Approved commercial uses

Cloned voices can be used for approved commercial projects when the speaker’s consent covers commercial use and the project complies with the Acceptable Use Policy. VoiceLuma displays the applicable license before generation so you know what is permitted.

Prohibited impersonation

Public-figure impersonation, cloning someone else’s voice without documented authorization, and any attempt to disguise a cloned voice as another person are strictly prohibited. Child and child-like voice cloning is prohibited.

Voice removal requests

A speaker can request review or removal of a voice modeled from them at any time. Send requests through the Contact page and the voice will be suspended pending review.

Data retention

Reference recordings and consent records are retained while the cloned voice exists on the platform, plus any period required to comply with legal obligations. Deleting a clone removes its associated voice model.

Watermarking

VoiceLuma applies audio watermarking whenever the connected speech model supports it, helping identify AI-generated audio.

Reporting abuse

If you believe a cloned voice was created without permission, report it through the Contact page. Reported voices are suspended while the report is investigated.