Captions on a non-English avatar clip: language hint and script_text

Inline captions on a Sume avatar video take four knobs: style, an optional font, a language hint and script_text. For a Spanish, French or German script, leave language at auto or pass the matching code, and pass your own script text so the burned words match what you wrote. The language field only tells speech-to-text what to expect; it never picks the style or font. Latin-script languages use the default Latin styles without any extra setting.

Source: Generate avatar video and Video captions.

What the fields do

The docs spell out the split: style picks the look and motion, design edits it, font picks the face, and language is only a speech-to-text hint. With script_text present, captions align your words to the audio instead of trusting a transcript alone.

A request

The inline form sits beside the script in the same talking-video body, and it is not a separate billed caption job. The soft-fail rule matters: if the caption stage fails, the avatar job can still succeed with a clean video_url and captions.status set to failed.

{
  "avatar_handle": "product_host",
  "script": "Hola, esta es nuestra nueva botella térmica.",
  "aspect_ratio": "9:16",
  "captions": {
    "enabled": true,
    "style": "slam",
    "language": "auto",
    "script_text": "Hola, esta es nuestra nueva botella térmica."
  }
}

Check the result like a reader

Open the captioned MP4 with the sound off and read it. Accents, inverted punctuation and long compound words are where a wrong transcript shows up. If one language looks wrong, re-run the caption step on the clean video with the standalone route rather than regenerating the avatar.

• Estimated duration above 60 seconds is rejected for inline captions.
• Preview stills are never captioned.
• Keep the clean video; re-caption without a new render.

Limits to remember

The avatar docs list no supported-language set for avatar speech, so test your language with a short script first and judge the voice and lip movement yourself. Captions are the part the docs specify precisely; the voice is the part to verify by ear.