Burn captions: pick one of script_text, words, cues or segments

A Sume video caption job takes its wording from one of four places, and the docs, read 2026-10-05, say you can send only one of script_text, words, cues and segments. The choice decides whether speech-to-text runs, so it is the first thing to settle.

The price does not depend on it: each accepted standalone caption job reserves and captures $0.20 for a video up to 60 seconds.

Which one when

Failure modes

• script_text can fail alignment with script_alignment_mismatch or script_alignment_failed. The suggested next action is simplify_script_text_or_omit.
• A speech-based job on a silent clip fails with caption_no_speech, and next_action points to use_overlay_captions.
• Korean text on a Latin style such as slam returns 400 caption_hangul_text_latin_style.

Authored cues on a silent clip

{
  "video_url": "https://media.sume.com/artifacts/example/silent.mp4",
  "style": "punch",
  "cues": [
    {"text": "New: free shipping", "start": 0.5, "end": 2.5},
    {"text": "Ends Sunday", "start": 2.5, "end": 4.5}
  ]
}

To change only the look afterwards, send source_caption_id and a new style. Sume reuses the source video and stored word timings, so speech-to-text does not run twice.