YOSUGA’s caption API draws the line

Field notes

The API accepts a media URL or upload, creates an asynchronous job, and returns transcript segments, caption cues, downloadable subtitle files, and optional filler or silence candidates. Its documented formatting layer includes twelve presets, from word-by-word karaoke highlighting to news-style captions, while uploads range from 20 MB direct posts to 2 GB signed uploads. The page states that completed jobs are billed by rounded-up media minutes, while failures and in-progress polling are free.