Caption Scribe by Jericho

Caption Scribe turns a video's captions into a document you would actually want to read.

Most caption exporters give you one line per caption cue. Auto-generated tracks come out worse: the scrolling effect encodes each phrase two or three times, so the export arrives duplicated and close to unusable. Caption Scribe merges the fragments into paragraphs and strips that repetition, breaking where the speaker pauses.

What you get:

• Markdown with YAML frontmatter recording the title, channel, duration, language, and whether the captions were written by a human or generated automatically.
• Clickable timestamps that jump straight back to the moment in the video.
• Chapter headings, taken from the video's own chapters.
• Plain text, SubRip (.srt) and WebVTT (.vtt) export as well.
• Control over timestamp density, paragraph length, frontmatter detail, and the filename pattern.

Nothing is collected and nothing is transmitted. No analytics, no telemetry, no remote code, and no server other than YouTube is ever contacted. Processing happens entirely in your browser, and the result reaches your clipboard or downloads folder only when you ask for it.

Where a limitation exists, the add-on tells you rather than papering over it. If cue timings can only be read approximately, subtitle export is disabled instead of writing a file with invented timings. If a requested language could not be honoured, the document reports the language it actually contains, never the one that was asked for.

The complete source ships inside the add-on, unminified and unbundled, exactly as written.