For security reasons, we cannot create a custom AI speaker without the consenting speaker's explicit recorded voice authorization. This is the person whose voice is being submitted for training.

Who can create an AI Speaker?

AI Speakers may only be created by or on behalf of Consenting Speakers who record their own training and authorization audio using their voice. Any other attempt to bypass the voice verification process will be considered a breach of our Terms of Service.

Please note that we currently do not accept any other authorization script for creating a custom AI Speaker than what is required by Descript.

For these reasons, we will not authorize, or provide exceptions to create a custom AI Speaker:

• For a deceased individual
• For a non-consenting speaker
• For individuals who are unable to record the training and authorization statement
• If the Voice training audio does not match the surrounding recorded audio (for Overdub use cases)
• For voice training audio originating from an AI or artificial source