A voice can feel personal in a way a stock image does not. That is why voice cloning deserves a higher bar than technical capability. Vois can keep a cloned voice and its production work on your own device, but it cannot decide whether you have permission to create that voice or whether listeners could be misled by it.
The useful path is also the responsible one: get clear consent, keep the scope of that consent visible, use the voice only for the approved purpose, and review the final audio before it reaches an audience. That protects the person behind the recording, the listener, and the project you are trying to build.
Get explicit permission before cloning
Permission should be informed, specific, and recorded. Explain that the audio will be used to create synthetic speech, describe the intended content and channels, and make clear whether the use is personal, internal, commercial, public, temporary, or ongoing.
A public recording is not a license. An old interview is not a blanket yes. Someone agreeing to narrate one internal training video has not agreed to become the voice of an advertising campaign. Keep the agreement in writing and make it easy to find when the project changes.
If the voice source changes their mind, stop future use while you clarify the situation. Consent is not a one-time technical checkbox.
Treat difficult cases as a reason to stop and ask
Some cases need more care than a quick project brief can provide. A deceased person's voice may involve family wishes, contracts, publicity rights, and cultural concerns. A public figure has not given up control of their voice by speaking in public. A minor may need legal consent from a guardian, but that may still not answer the ethical question of long-term use.
When the answer is unclear, do not turn uncertainty into a production deadline. Use a library voice or a designed synthetic voice instead, or obtain appropriate professional advice before you proceed.
Keep consent connected to the Vois project
Vois helps you keep production local, but a local clone is still a voice with a source and a scope of use. Before you add an approved sample, create a short record that identifies the voice source, the approved purpose, allowed channels, any date or territory limit, and the person responsible for questions.
Then keep the cloning decision and the project together. If the project changes from an internal course to a public campaign, or from one language to another, review whether the original permission still applies. If it does not, stop and obtain new permission or choose another voice.
For an original character or narrator that does not correspond to a real person, Pro Voice Design offers a different route. It creates a new voice from selected attributes rather than trying to recreate someone who already exists. That does not remove your responsibility to avoid deceptive uses, but it does avoid treating a real person's recording as raw material.
Tell listeners what they need to know
Disclosure should match the context and the risk of confusion. A short note in credits, a show description, a visible label, or an opening line can be enough when it clearly says that the recording uses synthetic narration.
Be especially clear when the content is commercial, educational, informational, or likely to be interpreted as a real person's statement. A synthetic voice that resembles a known individual can create a false impression even if the script does not name them.
Fiction can be different. A clearly fictional robot character or stylized audio drama may make the synthetic element obvious. Even there, straightforward transparency is usually easier than asking listeners to guess.
Uses that are off limits
Do not use synthetic voices for fraud, identity theft, false evidence, harassment, non-consensual intimate content, or impersonation intended to deceive. Do not make a public figure appear to say something they did not say. These are not creative edge cases. They can cause direct harm and undermine trust in real recordings.
A useful check is simple: would the person behind the voice, and the listener hearing it, understand what you made and feel that they were treated fairly? If the answer is no, do not publish it.
Put a human review gate before release
Synthetic voice projects need a review step just as much as they need a good render. Before you export or distribute audio, check the version against four questions:
- Consent: Do you have an explicit record that covers this speaker, project, and destination?
- Accuracy: Does the approved script say only what the source supports?
- Disclosure: Could a reasonable listener mistake this for a real person's recording, and if so, have you made the synthetic nature clear?
- Harm: Could the voice, script, or context unfairly injure, deceive, or expose someone?
The reviewer should listen to the actual Vois delivery asset, not only the source script. A correct script can still become a misleading or poor-quality recording through casting, pronunciation, or edit choices.
Put the technology to useful work
Consent-based synthetic voices can support audiobooks, accessibility, training, multilingual updates, games, and personal creative projects. Vois makes that work practical by keeping scripts and audio local while giving you a voice library, cloning where permission exists, and a full production workflow for review and export.
The standard is not whether a voice is technically convincing. It is whether the person whose voice is involved, the audience, and the project are all being treated with care. Explore voice cloning in Vois, review responsible production workflows, and Get started with a consent-first project.
The Vois Team