Voice & likeness rights
Voice and likeness rights is a practical umbrella term for legal interests associated with a recognizable person’s identity. Depending on the jurisdiction and use, publicity or personality rights, privacy, contract, and other rules may apply. Holding copyright in an audio or video recording does not necessarily authorize every use of the people depicted.
How it works
A review identifies whose voice or appearance is present, how it was recorded, the permissions obtained, and the proposed reuse. Model training, synthetic replication, advertising, and publication may raise different questions. The U.S. Copyright Office’s digital-replicas report discusses gaps and variations in existing protections; it is a report, not a blanket license or a current-law checklist for every jurisdiction.
Why it matters for licensing
Audio and video datasets can combine several layers of rights: the recording, scripts or music, and the interests of speakers or subjects. A licensing assessment should connect the intended technical use to the scope of actual agreements and any applicable legal requirements.
Example
Fictional example: A company has permission to publish a staff interview on its website. It considers using the recording to train a system that imitates the speaker’s voice. The earlier publication permission is reviewed rather than assumed to cover synthetic replication.
Limitations and misconceptions
Consent wording, employment context, collective agreements, and local law can affect the analysis. A blurred face or altered voice does not automatically resolve every concern. This entry does not claim that one release form works across all jurisdictions or AI uses.
Questions to ask
- Which people can be recognized, and what uses were actually authorized?
- Does the proposed use involve recognition, imitation, publication, or advertising?
- Which jurisdiction-specific rights and contractual limits need review?