What HeyGen documents about speech

HeyGen (developers.heygen.com) documents a voice clone that is instant from one recording or professional from twenty minutes or more and is then passed by id, plus video translation into thirty or more languages with lip-sync. From 1 source page, read 2026-09-22; no maker pays to be listed.

Inclusion rule. Fields are the same five for every model here; a field the vendor does not document is recorded as not documented rather than inferred from generated output. Order. Fixed field order, identical on every model page.

1Two grades of the same voice, with the threshold published

An instant clone comes from a single recording; a professional one asks for twenty minutes or more. Publishing both, with the boundary between them stated in minutes, tells a production what it has to book: a phone memo for a background part, a studio hour for a lead.

Once made, the clone is an id passed in the request, which makes the voice a reusable asset rather than a per-shot upload. That is the arrangement a season needs, and this entry states it in the same breath as the enrolment cost.

2Languages are counted where translation happens

The language figure on this platform is attached to translation rather than to generation: an existing video is translated and dubbed into thirty or more languages, with voice cloning and lip-sync. A production reading that as a list of languages it can perform in from scratch would be stretching the sentence.

Read as written it describes a route into another market that begins from a finished episode. That is the dubbing path with the mouth repaired afterwards, and it is a different plan from generating the performance in the target language to begin with.

3Speech as a step, and what that does to a schedule

Text becomes speech audio at its own endpoint, and the avatar is rendered from that. Pulling the two apart is useful: a line can be approved before a frame is rendered, and a re-render for a visual reason need not produce a new performance.

It also means nothing here arrives in one pass, so alignment is a service rather than a by-product. The register puts the lip-sync claim where the vendor makes it, beside translation, and leaves the mechanism unstated because the page read does not give one.

4This entry, one column at a time

Each of these stays inside a single field: what HeyGen puts there, what the wording settles, and what it leaves for a take to answer.

• Audio source — from a speech endpoint, then rendered.
• Voice source — a recording to clone from.
• Per-character binding — on a stored clone.
• Languages — a count, attached to translation.
• Lip-sync — named, alongside translation.

5Read against another entry

Each of these puts HeyGen beside one other entry on a column where the two land at opposite grades of answer.

• HeyGen and LTX Studio — on per-character binding.
• HeyGen and Synthesia — on languages.

6Where this model sits against the rest

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

Inclusion rule. Models whose vendor documents something about generated speech. A model with nothing published on the point is still listed, with the cell marked hollow. Order. Alphabetical by model name.

7Sources

Taken from the API quick start at developers.heygen.com on 2026-09-22. Every entry against the same five fields sits on the speech controls page; the rule behind each field is on how read. The entries either side of this one: Kling AI, LTX Studio.

• Voice source
  A 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 thresholdHeyGen, API quick start / recorded 2026-09-22
• Languages
  Video translation is documented for thirty or more languages, with voice cloning and lip-synccounted on the translation endpointHeyGen, API quick start / recorded 2026-09-22
• Audio source
  A text to speech endpoint turns a script into speech audio as a step of its ownan endpoint apart from the pictureHeyGen, API quick start / recorded 2026-09-22