AreaAudit

Which languages each generator says it supports, and where

No voice list, and a language claim about text inside the frame

PixVerse publishes no list of the languages its output can speak. Its V6 launch post claims multilingual text generation within frames, with accurate placement across English, Chinese and other languages, which is a claim about pixels rather than audio. As of 2026-09-22.

Four kinds of language claim, and which two this vendor makesSite languages describe the vendor own pages. Output languages describe generated documents. Voice languages describe the dialogue a video speaks. A claim about letters drawn inside the picture is a fourth thing, and it belongs in none of the three.Which system does the claim describe?The vendor own pagesSite column, filledThirteen declared barecodes on the host beingread.What the video speaksVoice column, emptyNothing published aboutwhat the output can say.Letters in the frameOutside all threecolumnsA launch post naming twolanguages, then others.The layer no subtitle or dub can repair afterwards
Fig. 1 Counting a frame-text claim as evidence about dialogue would be the same category error as counting a translated menu.
Four kinds of language claim, and which of them this vendor makes. Recorded 2026-09-22.
Kind of claimWhat it is aboutMade by this vendor
Site languagesThe vendor's own pagesYes, 13 declared codes
Output languagesGenerated scripts and breakdownsNo
Voice languagesThe dialogue a video speaksNo
Text inside the pictureLetters and characters the model drawsYes, in a launch post

Inclusion rule. The three register columns plus the one further kind of language claim this vendor's pages make, which belongs in none of them. Order. Register columns in fixed order, then the claim outside them.

1Text in a frame is a rendering problem, not a language one

Drawing legible letters inside a generated picture is hard for a different reason from speaking a language: it is about glyph shapes, spacing and placement rather than about pronunciation or grammar. A model can do one and not the other.

It matters for production all the same. A sign, a phone screen or a note written into a shot cannot be fixed by subtitles or by dubbing later, so a claim about frame text is a claim about how localisable the picture will be.

2Which languages, and how well, are both unstated

The post names English and Chinese and then says other languages. Nothing states which scripts are covered, and the scripts are where this capability breaks: a model trained mainly on Latin and Chinese characters is not evidence about Arabic or Devanagari.

So the claim is recorded as published and not converted into a list. It sits outside the three columns because putting it in any of them would let a reader take it as evidence about dialogue.

3A full site column and nothing about output

Thirteen declared site codes and no statement about the language of anything the product generates is the commonest profile in this register. Nine vendors declare a site list; five of those publish nothing in either output column.

On this row the gap is sharper than usual, because the frame-text claim shows the vendor does describe language behaviour when it has something to announce. What it has not described is the part a dubbed or subtitled release would depend on.

4Sources

Quoted from the PixVerse V6 launch post. The column itself is described on dialogue language, and every cell PixVerse fills is on its tool page.

5The same column, tool by tool

Voice languages, tool by toolOne column at a time, all 18 tools beside each other, PixVerse marked. A frame-text claim does not fill this column, so the cell reads hollow. Recorded 2026-09-22.Voice languages, tool by toolVoice languagesCapCutCapCut — Voice languages: No list publishedD-IDD-ID — Voice languages: 120+, count onlyFlikiFliki — Voice languages: 91 named, plus 117 dialectsHailuoHailuo — Voice languages: No list publishedHeyGenHeyGen — Voice languages: Two named lists, no totalHiggsfieldHiggsfield — Voice languages: No list publishedinvideo AIinvideo AI — Voice languages: No list publishedKling AIKling AI — Voice languages: Five, count onlyLTX StudioLTX Studio — Voice languages: No list publishedPikaPika — Voice languages: No list publishedPixVersePixVerse — Voice languages: No list publishedRask AIRask AI — Voice languages: 135+, count onlyRunwayRunway — Voice languages: No list publishedSceneMixerSceneMixer — Voice languages: 15, plus Cantonese for dialogueSynthesiaSynthesia — Voice languages: 143 namedVEEDVEED — Voice languages: 29 dub-to, 72 detectableVidnozVidnoz — Voice languages: 100+ and 140+, counts onlyViduVidu — Voice languages: No list published
Fig. 2 One column at a time, all 18 tools beside each other, PixVerse marked. A frame-text claim does not fill this column, so the cell reads hollow. Recorded 2026-09-22.
Every tool in the register on the voice languages column, with PixVerse marked. Recorded 2026-09-22.
ToolVoice languages
CapCutNo list published
D-ID120+, count only
Fliki91 named, plus 117 dialects
HailuoNo list published
HeyGenTwo named lists, no total
HiggsfieldNo list published
invideo AINo list published
Kling AIFive, count only
LTX StudioNo list published
PikaNo list published
PixVerseNo list published
Rask AI135+, count only
RunwayNo list published
SceneMixer15, plus Cantonese for dialogue
Synthesia143 named
VEED29 dub-to, 72 detectable
Vidnoz100+ and 140+, counts only
ViduNo list published

Inclusion rule. Tools with a public English site selling AI video generation or AI video dubbing. Site languages count hreflang alternates served from the host being read; x-default and alternates pointing at another host are not counted, for every vendor alike. Order. Alphabetical by tool name.

Neighbouring cells: Kling AI: dialogue language and Rask AI: dialogue language. All of them together: the cell index and the register table.

  • Voice languages
    No public list (as of 2026-09-12)no list publishedpixverse.ai / recorded 2026-09-22
  • Text drawn inside the picture
    “Multilingual text generation within frames is now supported, with accurate placement and style consistency across English, Chinese, and other languages”outside the three columnsPixVerse, V6 launch post / recorded 2026-09-22