Voice consent policy

Every cloned voice needs a paper trail.

This is the operating policy behind AudioPod's consent-first stance. Use it as a reference, share it with anyone whose voice you'd like to clone, and grab the template below.

Consent template

🎧 Listen to this page

Four requirements

What every voice clone on AudioPod has.

These aren't aspirational. They apply to every cloned voice — yours, your customers', everyone's.

You own or license the voice

Either it's your own voice, or you hold a written license from the person whose voice it is. No assumed consent, no 'I'll get it later.'

Consent is in writing

Email confirmation, signed PDF, or our downloadable template — something we can produce if a takedown dispute comes in.

Use is in scope

The intended use is allowed under our Responsible AI policy and within whatever the voice owner agreed to.

Audit trail kept

Every cloned-voice generation is logged with timestamp, prompt, and account context. Available on legitimate request.

What we accept and reject

Concrete examples on both sides.

Accepted

Your own voice, cloned for your own creative work
A friend or family member's voice, with their explicit written consent
A voice actor you've hired, under a licensing agreement
A historic / fictional character voice you've created from scratch
A voice you've licensed from a voice marketplace with clear synthetic-use terms

Rejected

Political figures, heads of state, or candidates for public office — at any scale, for any purpose
Living public figures who haven't given documented consent (celebrities, athletes, executives)
Anyone the cloner has reason to believe would object
Minors — voices of anyone under 18 require additional safeguards we don't currently support
Deceased individuals where their estate hasn't documented consent
Voices intended for impersonation-for-fraud, social engineering, or non-consensual sexual content

The flow

What happens from upload to takedown.

Upload

You upload a reference recording (5+ seconds, ideally 30+ for higher fidelity). You confirm consent at clone-creation time.

Automated checks

Public-figure detection scans the upload. Borderline uploads route to human review before clone is created. Clean uploads proceed to training.

Training

The clone is trained against your account only. Training data is encrypted at rest and not pooled into any general model without separate opt-in.

Generation

When you generate audio, the output carries provenance metadata identifying it as synthetic. The generation is logged against the clone.

Takedown (if needed)

If anyone files a misuse report — including the voice owner — we triage within 24 hours and complete verified takedowns within 72 hours.

Consent template

A signed paper trail you can use today.

A drop-in agreement between a voice owner and a voice user. Plain English, no legal jargon, ready to sign. Share with the voice owner, both sides sign, store the signed copy — done.

Voice Cloning Consent Agreement

Markdown source you can convert to PDF, sign, and store. Covers parties, grant of rights, permitted uses, territory, duration, revocation, compensation, and signatures. Templated — fill in the bracketed fields.

Download template (.md)

Starting point, not legal advice. For broadcast advertising, large publishing deals, or public-figure voices, talk to a media-and-entertainment lawyer in the relevant jurisdiction.

Report misuse

If something shouldn't have been made, tell us.

24-hour triage. 72-hour takedown.

Email [email protected] with the suspected AudioPod output, the original recording if you have it, and any usage context. A person reads every report.

Report misuse

Hold us to the policy.

This page exists so a voice owner, a journalist, or an enterprise legal team can read what we promise and check whether we deliver.

Get the template