Burned-in captions: ALL CAPS or as written? What each Sume style does

On Sume, the three Latin caption styles (slam, punch, tiktok-green) always burn ALL CAPS: the renderer upper-cases each word, so the case in your script_text, words or STT transcript does not survive. The Hangul-identity styles burn text exactly as written, with no case folding. Each standalone caption job costs $0.20 for a video up to 60 seconds.

Read on 2026-10-03: the Sume video-captions docs page and the caption renderer source it documents.

Where does the capitalization happen?

In slam, the word timeline is built with toUpperCase() before render, and the template also sets text-transform: uppercase on every word. punch and tiktok-green draw through a Remotion composition that sets textTransform: "uppercase" on the whole caption. So on those three styles, mixed-case input is a no-op for the look.

What do script_text, words and cues change?

They change the wording, not the case rule. script_text keeps STT timings and aligns your wording to them; words (word-level) and cues or segments (phrase cards) skip speech-to-text and burn your copy at your times. They are mutually exclusive. On a Latin style all of them still come out upper-case. On a Hangul identity they come out exactly as you typed them, including mixed-case product names or model numbers.

A silent clip has no speech to transcribe, so it fails as caption_no_speech unless you pass cues.

How do I get sentence case or Title Case?

Name a Hangul-identity style with English text. The request guard only refuses the opposite pairing, Hangul wording on a Latin style (caption_hangul_text_latin_style), so Latin text on weight-shift is accepted and not case-folded. The catch: those styles are authored for Korean, and I did not verify in code how well their faces draw English letters. Burn one test clip ($0.20) and look before you batch.

If you need a specific Latin display face in mixed case, the styles above do not offer it. Their font option accepts Hangul faces only, and naming one next to a Latin style returns caption_font_requires_hangul_style.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: caption-as-written-001" \
  -d '{
    "video_url": "https://example.com/clip.mp4",
    "style": "weight-shift",
    "script_text": "Meet the new iPhone case, in three colors."
  }'

Does the default style change anything?

Omitting style resolves to slam for Latin wording, so a default English job is ALL CAPS. It resolves to black-outline when the wording is mostly Hangul.