youtube-content

Use the youtube-transcript-api package to fetch transcripts by video ID or URL. The fetch_transcript.py script accepts watch links, youtu.be short links, shorts, embeds, and live URLs, returning JSON with segments, full text, and optional timestamps.

Fetch the transcript with the helper script using --text-only --timestamps, then transform it into a summary, chapters, thread, or blog post. Videos over 50K characters are chunked with overlap and summarized per chunk before merging.

Yes, pass a comma-separated language list with --language, such as tr,en, to set a fallback chain. If no matching language exists, retry without the flag to fetch any available transcript and note the actual language.

Fetching fails when transcripts are disabled by the video owner, the video is private or unavailable, or no transcript exists in the requested language. The script returns a JSON error message indicating which condition occurred.

Transcripts can be transformed into timestamped chapters, 5-10 sentence summaries, chapter summaries, Twitter/X threads under 280 characters per post, full blog posts with sections, and notable quotes with timestamps.