Voices you may dub
A dub speaks in the voices of the people in the video, so every project starts with the right to use them. Here is what Evocal asks, what it records, and what it does not allow.
Whose voice you may dub
Your own voice, or the voices of people who gave you their written permission. Public figures are no exception: their voices need their permission like anyone else's.
The confirmation for each file
Before the first dub of a project, you confirm: “I have the right to use every voice in this file: it is my own, or I have written permission from each person who speaks.” The answer is recorded with the file's SHA-256 fingerprint, so it belongs to that exact video and to no other. A new file asks again.
Marked as AI
Every dub is exported with AI marking: provenance information in the file and a watermark in the audio. Our terms ask you to tell your audience that the dub was made with AI, for example in the video's description.
What is not allowed
- Passing yourself off as someone else
- Fraud, scams and automated calls
- Sexual content with a real person's voice
- Political or election disinformation
- Harassment and threats
- Voices of public figures or other people without permission
- Copyright infringement
Breaking these rules leads to the content being removed and the account suspended, or banned in serious cases. The full rules are in the Acceptable Use Policy and the voice guidelines.
Your own voice and Evocal's research
Your own voice, and only the one you mark as yours in a project, may be used to improve Evocal's transcription, translation and dubbing. It is never sold and never used to build a model of one specific person's voice. It is kept for up to three years, and it leaves the archive and future training if you close your account. In the EU, the EEA and the UK this rests on legitimate interest instead, and you can object at any time. The details are in the study terms.