Acceptable use policy

Separating voices is useful for editing, transcription, accessibility, and research. The same capability is open to misuse, mostly around recording people who did not agree to be recorded; don't do that, and the rest of this page follows from it.

Only upload audio you have the right to process. Recording laws differ by country and by state, and some require every participant's consent; meeting that obligation is yours, not ours.

• Do not upload covert recordings, or audio obtained by intercepting a conversation.
• Do not use the service to isolate a person's voice for surveillance, stalking, or harassment.
• Do not use output to impersonate someone, including as source material for voice cloning without that person's consent.

• No audio depicting the sexual abuse or exploitation of children, in any form.
• No content whose purpose is to threaten, defame, or incite violence against a person or group.
• Do not upload audio you do not have the copyright or license to process.

• Do not try to circumvent rate limits, quotas, or billing, including by creating accounts to collect the signup credit repeatedly.
• Do not attempt to access another account's jobs, keys, or audio, or to probe the service for vulnerabilities without asking us first. If you find one, tell us.
• Do not resell the API as your own without an agreement. Talk to us about partnership instead.
• Do not use the service to extract our models, or to build a directly competing model from its outputs.

We do not routinely review uploaded audio, so enforcement is mostly reactive: we act on reports and on abuse we detect operationally. Where we find a breach we may suspend or close the account, and where the law requires it we may report the matter; we will tell you why, unless telling you would itself be unlawful.

To report misuse, or to ask whether an intended use is acceptable before you build on it, get in touch.