HeyGen: a clone gets an id, and the id is the character
A clone is enrolled once and afterwards passed as a voice id. That makes the voice an object with a name, created before any shot and reused on every one, which is the arrangement a long shoot needs. As of 2026-09-22.
| What the documentation settles | What it leaves to a take |
|---|---|
| A clone is instant from one recording, or professional from twenty minutes or more | Which grade a production should enrol for a recurring character |
| The clone is then passed as a voice id in the request | Whether an enrolled clone can be revised without a new id |
| Video translation is documented for thirty or more languages, with cloning and lip-sync | Whether the same clone id carries across those languages |
Inclusion rule. Only statements found on the vendor page named in the caption, and only ones bearing on this one field. Nothing here is filled in from generated output. Order. In the order the statements arrive on the vendor's page.
1Enrol once, and the drift has one chance to happen
A voice built once is wrong once if it is wrong at all. That is the argument for enrolment over a clip supplied per call: the quality of the source recording is a decision taken deliberately at the start rather than repeated, imperfectly, forty times.
It also front-loads the work. Twenty minutes of usable audio for the professional grade has to exist before a character can speak, so the casting calendar sits in front of the production calendar rather than beside it.
2A clone that travels into thirty languages is a bigger commitment
Because translation is documented with cloning, the same enrolled voice can appear in markets the performer never agreed to in conversation. The mechanics make that easy; the release form is the only thing that makes it lawful, and nothing on the page discusses it.
For this register the relevant part is that the id is the unit of reuse. Whatever a production negotiates has to be attached to an id in its own records, because the platform will happily use it anywhere.
3The others where a voice exists before any shot does
Three more entries let a voice be created, named and reused. Two of them bind it to a character object instead of to a voice object, which changes who has to remember the pairing.
- Kling AI — on the element.
- LTX Studio — on the character element.
- Runway — on a stored voice.
- 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
- LanguagesVideo translation is documented for thirty or more languages, with voice cloning and lip-synccounted on the translation endpoint
4Sources
Read from the API quick start at developers.heygen.com on 2026-09-22. The same column across every entry is on per-character binding; everything this vendor publishes about speech is on HeyGen. What counts as documented is on how read.