Acceptable Use Policy

This policy applies to everything you do with Voice Buddy — the console, the API, and anything you generate with either. It forms part of the Terms of Service, and breaching it is a breach of those terms.

It applies to what you generate, what you upload, and what you do with the result after you download it. Taking a file off the platform does not take it outside this policy.

Being straight about the mechanism, because it changes where the responsibility sits:

• Nothing screens your text or audio before it renders. There is no moderation model, keyword filter or classifier between what you submit and the speech engine. We find out about a breach when somebody tells us.
• Generated audio carries no watermark. A finished file contains no marker identifying it as synthetic or as ours. Nothing downstream will catch a misuse that you decide to commit.
• We do keep records — every render is stored with the script, the voice, the settings, the workspace and the user, and API calls are logged with the IP address. So a misuse is traceable after the fact, which is exactly the wrong order for the person it was aimed at.

You are therefore the control. The rules below are not a formality; they are the only thing standing between this tool and the people it could be used against.

Do not clone, synthesise or convert into a voice unless it is your own or you hold the speaker’s express, informed, documented consent for the uses you intend. That includes:

• A public figure, politician, executive, celebrity or broadcaster, however much of their voice is already public.
• A colleague, client, family member or ex, however well you know them.
• A deceased person, without the consent of whoever holds the rights to their likeness and voice.
• A child, without the consent of a parent or guardian.
• A voice actor’s performance, beyond the licence they granted for that performance. A recording you paid for is not consent to clone the person who made it.

The Voice Cloning and Biometric Data Notice sets out what that consent has to cover and how a speaker asks for their voice to be erased.

Impersonation and deception

• Passing off synthetic speech as a real person, living or dead, in a way likely to make someone believe the person actually said it.
• Impersonating a bank, a government body, an employer, a police force, a hospital, a delivery service or a support desk.
• Defeating voice authentication or a voiceprint security system — including your own bank’s, and including “just to see if it works”.
• Fabricating evidence, a confession, an interview, a recorded call or a news report.

Fraud and financial harm

• Scam calls, robocalls, vishing, romance and investment scams, and the “grandchild in trouble” call in every language it exists in.
• Authorising or instructing a payment or transfer in someone else’s voice.
• Fake endorsements or testimonials, and advertising that puts words in the mouth of a person or brand who never agreed to them.

Harassment, abuse and sexual content

• Content that harasses, bullies, threatens, degrades or intimidates a person, or that is designed to be sent to a person to distress them.
• Sexual or intimate content in a real person’s voice without their explicit consent to that specific use.
• Any sexual content involving a minor, or any content that sexualises a child’s voice. This is reported to the authorities, not warned about.
• Hate speech, or content inciting violence or discrimination against people on the basis of a protected characteristic.

Political and civic deception

• Synthetic speech attributed to a candidate, official or party that they did not say, in any context connected with an election, referendum or public consultation.
• Content about when, where or how to vote, voter eligibility, or the integrity of an election, that is false or misleading.
• Robocalls or messages to voters in a synthetic voice, whether or not the voice imitates anyone specific, except where you comply with every applicable election and telecommunications law and disclose that the voice is synthetic. Jurisdiction-specific disclosure rules are your responsibility. A number of US states regulate synthetic media in political advertising, calls placed to US telephone numbers are subject to federal rules on artificial and prerecorded voices as well, and several of those regimes prescribe the wording, the placement or the duration of the disclosure rather than leaving its form to you. We do not track which of them apply to your campaign, and this policy is not legal advice — settle that question before you publish or dial, not after.
• Impersonating an emergency service or public health authority, or generating false emergency, disaster or public-safety announcements.

Misinformation with physical consequences

• Medical, legal or financial advice presented as coming from a qualified professional who did not give it.
• Instructions for weapons, explosives or attacks on people or infrastructure.

Rights of others

• Uploading audio, music, performances or scripts you do not have the rights to.
• Ignoring the licence on a sound effect taken from the Freesound library — several require attribution, and the licence and author are recorded next to every result for that reason.
• Publishing content that is defamatory, or that breaches someone’s privacy or confidence.

Platform integrity

• Sharing API keys outside your organisation, reselling raw access to the API, or operating a service that is a thin wrapper over it, unless we have agreed to it in writing.
• Creating multiple accounts or workspaces to get around plan allowances, or using the Free plan’s allowance for commercial work.
• Probing, scanning or load-testing the service, or attempting to reach another workspace’s data. Coordinated security testing is welcome — arrange it first by emailing [email protected], marked for the attention of the security team, and agree the scope and the window before you start. Test against your own workspace, your own voices and your own recordings, never anyone else’s.
• Coming back after a termination under a new account.

Where a listener could reasonably think they are hearing a real person’s recorded voice and it matters — advertising, journalism, political communication, customer service, anything financial — say that the voice is synthetic. A short spoken or on-screen disclosure is usually enough, and in several jurisdictions it is required.

This is a floor, not a compliance opinion. What the law demands where you publish is yours to determine.

If you believe your voice — or a voice you hold rights in — has been cloned on Voice Buddy without consent, or you have encountered content made here that breaches this policy, email [email protected], marked for the attention of the abuse team. Tell us what you can: a link or file, when you encountered it, and how you are connected to the voice. You do not need an account with us to report something.

Reports about a person’s own voice are treated as urgent. We acknowledge a report within two business days — business days being Monday to Friday, 8:00 AM to 6:00 PM US Central — and can disable a voice or suspend a workspace while we look into it.

Depending on what happened, we may:

• Ask you to stop, and give you a chance to put it right.
• Remove a voice, a render or a project.
• Suspend or terminate the workspace or the account.
• Report the matter to law enforcement or a regulator, and preserve the associated records.

For cloning a voice without consent, sexual content involving a real person or a minor, fraud, or impersonation of an emergency or financial institution, we will act immediately and without notice. Terminating a paid workspace for a breach does not entitle you to a refund of the current period; see the terms.

Appeals. If you think a decision was wrong, email [email protected] within 30 days of it, marked for the attention of the appeals team. Name the workspace and the decision, and include whatever we did not have in front of us — a consent record, a licence, the context a clip was taken from. We will look again and tell you what we conclude. Two things worth knowing before you start: a suspension normally stays in place while an appeal is open, and deletion is not reversible. Removing a voice or a workspace deletes the stored audio itself rather than hiding it, so an appeal we uphold can restore your access but not your files.