Draft — not yet reviewed by counsel.
Acceptable use policy
Last updated October 29, 2026
Voice cloning is useful for accessibility, creative work, games, podcasts and personal tools. It can also hurt people. Lyrevox is built on Voicebox, and this policy follows the spirit of Voicebox’s Responsible Use guidelines: respect other people’s voices. It applies to everything you do with Lyrevox, in the browser and in the desktop app.
Consent comes first
Only clone voices you have the right to use:
- your own voice;
- a voice whose speaker has given you explicit permission;
- voice material that is licensed, in the public domain, or otherwise legally available for this use.
We can’t verify who a voice belongs to. That responsibility is yours.
Not allowed
- Impersonating a real person without their permission.
- Fraud, scams, phishing, social engineering, or getting past voice-based authentication.
- Harassment, threats, intimidation, or non-consensual sexual content.
- Any sexual content involving minors.
- Misleading political, legal, financial, medical or emergency communications.
- Commercial use of a person’s voice without the legal right to do so.
- Removing or tampering with the AI watermark described below.
- Attacking or overloading the service, trying to reach other customers’ studios or data, getting around usage limits, or reselling or sharing access to your account.
Disclose synthetic audio
If you publish or distribute audio made with Lyrevox, say that it is AI-generated wherever the law, a platform’s rules or your audience would expect it.
Watermark
Audio generated with the Chatterbox and Chatterbox Turbo engines carries Resemble AI’s Perth watermark: an inaudible signal that lets the audio be identified as AI-generated. Lyrevox keeps this watermark switched on. Audio from Qwen3-TTS and Kokoro is not watermarked.
Enforcement
If we find or are told about use that breaks this policy, we may remove access to content, suspend the studio or close the account, and we may report illegal activity to the authorities. To report misuse, contact [contact email to be added before launch].