youtube-subs

To extract clean plain-text transcripts from YouTube subtitles, this Skill fetches VTT files via yt-dlp and strips timestamps, deduplicates captions, and reflows text. This produces readable transcripts specifically formatted for large language model ingestion.

Yes, you can download subtitles from a YouTube playlist or channel at once. The fetch script supports automated batch subtitle downloading for multiple videos, generating clean transcripts and an index.tsv file containing metadata like duration and view counts.

Yes, you need yt-dlp, Python 3, and jq installed in a local Bash environment to download YouTube transcripts. These dependencies execute the subtitle fetching and text cleaning scripts required to output the final plain-text files.

YouTube subtitle transcripts contain duplicate lines due to overlapping VTT caption timestamps. The transcript cleaning process strips these timestamps, deduplicates the repeated text, and reflows the remaining characters into continuous plain-text.

The best way to summarize learning lectures from YouTube video transcripts is to batch download the subtitles as clean plain-text. This Skill outputs deduplicated text and an index.tsv file, making the content immediately ready for AI summarization.

When extracting transcripts from YouTube videos, the Skill generates an index.tsv file containing metadata for each video. This index includes the video ID, title, duration, upload date, and view counts to help organize large-scale learning batches.