youtube-transcript

To download YouTube captions and convert VTT to a Markdown transcript, the Skill fetches available subtitle tracks, downloads the VTT file, and uses a local conversion script to produce timestamped text. It outputs a clean Markdown file with per-line timestamps and the source video link.

Subtitle discovery works by listing all available subtitle languages and tracks before downloading the YouTube video transcript. This ensures you can select the best available option, prioritizing human-created subtitles over auto-generated captions.

Yes, you need yt-dlp installed to download YouTube subtitles, as the Skill relies on it to fetch available tracks. The tool resolves yt-dlp availability via the system PATH or a shared virtual environment.

If a YouTube video has no manual subtitles, the Skill falls back to downloading auto-generated captions to produce the transcript. If no subtitles exist at all, it only uses Whisper transcription after confirming with the user.

Yes, you can extract quotes and searchable text from a YouTube interview by downloading its captions and converting them into a timestamped Markdown transcript. This format makes it easy to search content and pull specific quotations for notes.

Yes, a limitation exists when transcribing YouTube videos without available subtitles, as the Skill cannot automatically generate text from audio alone. It requires explicit user confirmation to invoke Whisper as a fallback for auto-generated transcription.