AI Voice & Cloning Policy

1. Why this policy exists

AI voice technology can do a lot of good. It helps creators publish faster, gives people who have lost their voice a way to be heard, makes content accessible in more languages, and lets small businesses sound professional without a studio. The same technology can also be misused to scam families, mislead voters or humiliate people. We want Speechma to stay on the right side of that line.

This policy applies to all Speechma tools that reproduce, transform or imitate a human voice, especially Voice Cloning, Voice Changer, AI Voice Cover and AI Dubbing. It is part of our Terms of Service and works alongside our Acceptable Use Policy.

2. Our principles

• Consent. Every real human voice belongs to a person. You need that person’s permission to clone or transform it.
• Honesty. Synthetic audio should not be passed off as a real recording of a real person in a way that misleads listeners.
• No harm. Voice tools must never be used for fraud, harassment, sexual exploitation, election interference or other harm.
• Privacy. Voice samples are processed only to produce the audio you request and are not stored on Speechma servers after processing. We do not build voiceprints or use samples to identify anyone.
• Responsiveness. When misuse is reported, we act quickly and take it seriously.

3. Consent rules for voice cloning and voice changing

You may clone, transform or create a cover with a human voice only if one of the following is true:

• It is your own voice.
• You have the speaker’s clear, informed permission for the specific kind of use you plan. For example, a voice actor who has agreed in writing that you may create an AI version of their voice for your audiobook series, or a colleague who has agreed to have their voice used for your company’s training videos.
• You have legal authority to give permission on the speaker’s behalf, such as a parent or guardian for a minor’s voice used in a family project, or an estate that controls a deceased person’s voice and publicity rights.

Good consent is specific (it covers what you will create and where it will be published), informed (the person understands it will be an AI-generated version of their voice), freely given and revocable for future uses. For commercial projects, we strongly recommend getting consent in writing and keeping a copy.

Some things are not consent: having a recording of someone, finding their voice on YouTube or a podcast, the person being a public figure, the use being “just a joke,” or a contract that never mentioned AI voice replication. Several U.S. states, including Tennessee under the ELVIS Act and California, protect individuals’ voices and likenesses by law, and similar rights exist in many other countries.

4. No impersonation of celebrities, politicians or real people

Speechma does not offer celebrity voices, and you may not use our tools to recreate or imitate the voice of any celebrity, actor, musician, athlete, influencer, journalist, politician, candidate, government official or other real person without their consent. This includes:

• “sound-alike” clones made from interviews, songs, speeches or videos;
• AI voice covers that present a song as if it were sung by a famous artist;
• parody or satire that a reasonable listener could mistake for a real statement; and
• synthetic audio of political figures, at any time and especially around elections.

Need a distinctive voice for a character, trailer or narrator? Use our 1,700+ ready-made voices, AI voices such as narrator, trailer and documentary styles, or Voice Design to create a brand-new voice from a written description. These are original synthetic voices that do not belong to any real person.

5. Disclosure and labeling of AI-generated audio

We recommend that you tell your audience when audio is AI-generated, and in many situations you are required to. A simple, clear label works best, for example:

• “Narration created with an AI voice.”
• “This video uses AI dubbing. The original speaker’s words were translated and re-voiced.”
• “Voice: AI clone of [name], used with permission.”

Always disclose when content uses a cloned or transformed voice of a real person, when it could be mistaken for a genuine recording, when it is used in advertising or endorsements, and when it appears in news-style, political or informational content.

Follow platform rules. Major platforms, including YouTube, TikTok and Meta, require creators to label realistic synthetic or altered content and provide tools for doing so. Advertising networks and some U.S. state laws also require disclosures for AI-generated content in political ads and automated calls. You are responsible for complying with the rules that apply to where and how you publish.

Keep labels with the content. If you edit, repost or license Outputs, keep the disclosure in the description, credits or metadata so it travels with the audio.

6. Uses that are never allowed

Regardless of consent, you may never use Speechma voice tools to:

• commit fraud or scams, including fake emergency or “family in trouble” calls, CEO or bank impersonation, or payment authorization by voice;
• bypass voice authentication or identity verification systems;
• create election or civic misinformation, or imitate candidates or officials;
• harass, threaten, bully, defame or humiliate anyone;
• create sexual content involving a real person without their explicit consent, or any content involving minors in a sexual context;
• make robocalls or prerecorded calls that violate the TCPA or similar laws; or
• create content that is otherwise illegal or violates our Acceptable Use Policy.

7. Special care: minors, deceased people and sensitive contexts

• Minors. Only a parent or legal guardian may consent to the use of a child’s voice, and only for appropriate purposes such as family projects. Never publish content that could put a child at risk.
• People who have died. Recreating the voice of someone who has passed away can be meaningful for families, but it can also cause real distress. Only do so with the permission of the person’s family or estate, and never in a way that misrepresents what they said or believed. In many places, publicity rights continue after death.
• Health, legal and financial content. Do not use a real person’s voice to give medical, legal or financial advice they did not give.
• Accessibility uses. We support people using voice cloning to preserve or restore their own voice. If you are helping someone with a speech condition, make sure they, or their authorized representative, agree to how their voice will be used.

8. How we help prevent misuse

We design the Service to reduce the risk of abuse while keeping it free and simple:

• Voice tools are presented with clear reminders that you must have the right to use the voice you upload.
• Every generation requires a security code, and we may apply usage limits, which makes large-scale automated abuse harder.
• Automated access, scraping and limit circumvention are prohibited and actively blocked.
• We do not offer a library of celebrity or public-figure voices.
• We review reports of misuse and act on them, and we cooperate with law enforcement and platforms when appropriate.

Because we do not store uploads or generated results, we cannot search for or recall content after it has been created. That is why responsible use by each person, and reports from the public, are so important.

9. Reporting misuse and how we respond

If you believe your voice, or the voice of someone you represent, was cloned or imitated with Speechma without permission, or you have found AI audio created with Speechma that is being used to deceive or harm, please email contact@speechma.com with the subject “Voice Misuse Report.” Helpful details include:

• links to, or copies of, the content and where it was published;
• whose voice is involved and your relationship to that person;
• why you believe the content was made with Speechma; and
• how to reach you.

What we do next. We acknowledge reports as quickly as we can and prioritize those involving fraud, threats, minors, sexual content or elections. Depending on the facts, we may block related access, suspend Speechma API accounts, preserve available information, refer the matter to law enforcement, and provide information to help you request removal from the platform where the content was published. Most platforms have their own reporting tools for impersonation and synthetic media, and using them together with your report to us is usually the fastest way to get content taken down.

If you are facing an urgent threat or have lost money to a scam, contact local law enforcement right away. In the U.S., you can also report to the FTC at reportfraud.ftc.gov and to the FBI’s Internet Crime Complaint Center at ic3.gov.