Whether any vendor says whose voice may be used
Cloning is documented as a feature by four entries and as a risk by one. That one names United States digital replica statutes and treats an identifiable voice as a likeness question, which rules a method out rather than describing one. As of 2026-09-12.
| Model | What is documented | Detail |
|---|---|---|
| HeyGen | Cloning documented, permission not addressed | Instant, or twenty minutes and up |
| PixVerse | Custom voices from a supplied sample | Permission not addressed |
| Runway | A voice from a sample, or from a description | Permission not addressed |
| SceneMixer | Cloning an identifiable voice named as a likeness issue | Digital replica statutes named |
| Synthesia | Cloning with its own language range | Permission not addressed |
Inclusion rule. Entries whose documentation either describes building a voice from a recording or says something about permission to do so. An entry that never mentions cloning does not earn a row. Order. Alphabetical by model name.
1A restriction is documentation too
Most of this register records what a vendor says a production may do. One entry records what it says a production may not, and that sentence names two permitted routes in passing: a preset library and the user's own recordings.
It is the only place here where a boundary is drawn at all. The framing is unusual enough to note: a compliance checklist rather than a product page, written to constrain rather than to sell.
2The mechanics are easy and the permission is not on the page
An instant clone from one recording is a few minutes of work, and it travels: one platform documents cloning alongside translation into thirty or more languages. The voice being carried into thirty markets belongs to somebody, and the release form is the only thing that makes it lawful.
No API quick start contains a release form, and this register does not expect one to. Recording the silence is the useful part, because a team reading only the mechanics will not be prompted to ask.
3A described voice sidesteps the question entirely
One entry lets a voice be asked for in words rather than built from a recording, with a published minimum of twenty characters. There is no identifiable person behind a sentence about a voice, which makes it the cleanest casting route here for a minor character.
It is also the least controllable. A description produces something plausible and not something chosen, which is a trade worth making for a character with three lines and not for a lead.
- What the vendor rules outCloning an identifiable voice is described as a likeness issue, with US digital-replica statutes named
- Voice sourceThe guide tells users to use a preset voice library or their own recordingstwo routes named
- Voice sourceA voice clone is instant from one recording or professional from twenty minutes or more, and is then passed as a voice idtwo enrolment routes with a threshold
- Voice sourceA custom voice can be built from an audio sample between 10 seconds and 5 minutes long and at most 10 MBa published sample window
- Voice sourceVoice cloning carries a language list of its own, running from Afrikaans to Zulu and described as the same list used for standalone cloninga cloned voice with its own reach
- Voice sourceText to speech supports both built-in voices and custom voices created from user-provided sample audiocatalogue or supplied sample
4Sources
Every line is read from the vendor documentation linked on the speech controls page, checked 2026-09-12. What counts as documented is on how read. Nearby: A second control, What kind of page, Formats and file sizes.