Languages: which ones are named for dialogue
Languages records which languages a vendor names for spoken dialogue, not which languages its interface runs in. Two entries publish names. Three publish a total with nothing named. Three claim breadth without naming anything. The remaining nine say nothing at all. As of 2026-09-12.
| Model | What is published for dialogue | The published wording behind the cell |
|---|---|---|
| D-ID | A language field, no list | An optional language sits beside the chosen voice |
| Hedra | A claim with no names | Full multi-language support |
| HeyGen | A count, attached to translation | Thirty or more languages, with cloning and lip-sync |
| Kling AI | A count, no names | Native audio in five languages, on VIDEO 3.0 |
| LTX Studio | Not documented by the vendor | Nothing published about spoken languages |
| Luma Ray | Not documented by the vendor | Nothing published about spoken languages |
| MiniMax | Not documented by the vendor | Speech is documented without a language list |
| PixVerse | Multiple, none of them named | Multiple languages and audio types, including singing |
| Runway | Not documented by the vendor | A multilingual model option, with no language named |
| SceneMixer | A named list, set per project | Fifteen dialogue languages, with Cantonese added for dialogue only |
| Sora 2 | Not documented by the vendor | Dialogue guidance that never mentions a language |
| sync-3 | A count, ninety-five or more | The same coverage as the models before it |
| Synthesia | A catalogue with codes | A formal name, a native name and a code, per voice |
| Veo | Not documented by the vendor | Audio is documented; the language it speaks is not |
| Vidu | Not documented by the vendor | Three audio tracks documented, no language named |
| Wan 3.0 | Not documented by the vendor | Audio is documented; the language it speaks is not |
| Wan2.2-S2V | Not documented by the vendor | Audio-driven, so the language is whatever was recorded |
Inclusion rule. Models whose documentation names or counts the languages of the spoken performance. Interface languages, subtitle exports and prompt languages are not this field and do not earn a row. Order. Alphabetical by model name.
1A count is not something a schedule can be planned against
Five languages and a named list of fifteen are not the same kind of statement, even though both look like coverage in a feature table. A count tells a production that the capability exists somewhere inside it. A list tells a production whether the market it is shipping to next quarter is in or out, which is the question being asked.
The field keeps the difference visible by recording the wording rather than a number. Where a vendor counts, the cell says it counts, and the gap between a count and a list is left where the vendor put it instead of being filled in by inference.
2Interface languages and dialogue languages are separate lists
Product pages tend to carry a language list that belongs to the menus, the documentation or the prompt parser, and that list is longer than the one the actors speak. Reading the two as one is the most common way this field is misread, and it converts into a market decision that the documentation never supported.
So only a language named for the spoken performance is recorded here. Where a vendor publishes a list without saying which of the two it describes, the cell records the ambiguity rather than resolving it in the vendor's favour.
3Named is still not the same as usable in that market
One entry states that the line is performed in the chosen language while the shot renders, with no dubbing pass afterwards and therefore nothing to synchronise. That is the strongest claim in the column, and it is a claim about mechanism rather than about how the delivery sounds to someone who speaks the language at home.
Register, accent and idiom are outside what any page can settle, and no vendor read for these notes documents them. A named language moves a market from impossible to testable; whether it moves it to shippable is decided by a listener rather than by this field.
4One entry at a time on this column
A cell gets a page of its own where the vendor says something specific in it, or where its silence is unusual among the entries answering the same way. The remaining cells are left in the table above, because a page repeating one short phrase would be worse than a row carrying it.
Entries that publish names rather than a total:
- SceneMixer — a named list, set per project.
- Synthesia — a catalogue with codes.
Entries that publish a total with no names beside it:
- HeyGen — a count, attached to translation.
- Kling AI — a count, no names.
- sync-3 — a count, ninety-five or more.
Entries that claim more than one language without naming any:
- D-ID — a language field, no list.
- Hedra — a claim with no names.
- PixVerse — multiple, none of them named.
Entries that never say which language a performance is delivered in:
- Runway — not documented by the vendor.
- Sora 2 — not documented by the vendor.
- Wan 3.0 — not documented by the vendor.
- Wan2.2-S2V — not documented by the vendor.
- LanguagesDialogue is set per project to any of 15 languages, with Cantonese a 16th for dialogue onlya named list of 15
- Lip-syncThe vendor states there is no separate dubbing step and no lip-sync patch afterwardsstated as unnecessary rather than as a feature
- Audio sourceNative audio on VIDEO 3.0 in five languagesgenerated with the picture
- Languagessync-3 supports 95 or more languages, described as the same coverage as the models before itpublished as a count
- LanguagesEach voice is listed with its formal and native language name, a language code, a gender, a name and a voice idpublished as a catalogue with codes
- LanguagesVideo translation is documented for thirty or more languages, with voice cloning and lip-synccounted on the translation endpoint
- LanguagesMultiple languages and audio types are supported, including speech, singing, and advertisementsmultiple, none of them named
- LanguagesDescribed as text, image and audio to video with full multi-language support, and a maximum duration of ten minutesa claim carrying no list
- LanguagesNot documented by the vendor (as of 2026-09-22)neither a list nor a count
- LanguagesNot documented by the vendor (as of 2026-09-22)nothing published about the spoken performance
5Sources
Each cell is read from the vendor page it links to, checked 2026-09-12. The fields sit side by side on the speech table, and what counts as documented is set out on how read. The other fields: Lip-sync, Audio source. All of them: the field notes.