Dialogue language and video tier

Work in your own language covers the four language settings a project has.
Where dialogue language is set
In step 1, before the script is read. It decides the language the characters speak in every segment; the storyboard copies lines as written, and the setting is fixed once the script is read; for another language, start a new project with the same script. Changing the video tier does not change the language; it only changes who renders the speech.
How the tiers compare
| Language group | Default tier (Wan 3.0 · 480P) | Other tiers |
|---|---|---|
| German, Ukrainian | Flagged: an occasional mispronounced word (verb or case endings) | No tier is marked better; the same care applies |
| The other dialogue languages | Not flagged | No tier is marked better |
The picture is the same across the Wan, Seedance and H3 families: none of them is flagged as the fix for German or Ukrainian, so switching tiers for pronunciation alone is not a good use of credits.
What actually helps
Shorter sentences. A short line leaves less room for a slip than a long one. Regenerate the segment. A slip is not fixed; the same line can come out right the next time, and every version is kept. Check names. A foreign name in the middle of a line is where most slips happen; spell it the way it is pronounced in the dialogue language, or give the character a name native to that language in step 1.
A warning you may see before generating
If a segment’s lines are in a different language from the project’s dialogue setting — because you edited them by hand — SceneMixer asks you to confirm before rendering. Confirm to render as written, or fix the lines first.
FAQ
Does a more expensive tier pronounce German better?
No tier on SceneMixer is marked as better for German or Ukrainian. Use shorter lines and regenerate.
Can characters speak different languages in one episode?
A SceneMixer project has one dialogue language; lines you write in another language are spoken as written after a confirmation.
Why did a Japanese name come out with the wrong reading?
Names are the most common slip. Write the name the way it is pronounced, or pick a name with one reading.
Does regenerating cost the full price?
Yes; every render on SceneMixer is a new charge. If one fails, its credits come back or are held until support checks it.
Is the voice the same across tiers?
On SceneMixer, every tier uses the character’s voice setting, up to the tier’s limit on voice clips: at most 3 on the H3 tiers and 5 on the others, and no more than about 14 seconds of voice samples in total, which in practice is often three voices. A character beyond that limit speaks from the written voice description.
Hear your lines in your language
Set the dialogue language in step 1, keep lines short, and generate a segment on the default tier.
Updated November 15, 2026