Downloads and processes YouTube video transcripts, captions, and subtitles into clean text or VTT formats.
This skill provides a robust automated workflow for extracting text content from YouTube videos using yt-dlp and OpenAI's Whisper. It intelligently navigates the retrieval process by first checking for high-quality manual subtitles, falling back to auto-generated captions, and offering a sophisticated AI-powered transcription path using Whisper if no native subtitles exist. The skill excels at post-processing, featuring a Python-based deduplication engine that converts messy, overlapping WebVTT files into clean, readable plain text while preserving the original narrative flow.
主な機能
01Flexible output formats including both raw VTT and sanitized plain text files
02Smart post-processing to deduplicate overlapping lines and remove VTT timestamps
03Automatic detection and prioritized download of manual vs. auto-generated subtitles
041 GitHub stars
05AI-powered transcription fallback using OpenAI Whisper for videos without subtitles
06Automated environment setup and dependency checking for yt-dlp and Whisper
ユースケース
01Extracting specific quotes and content from YouTube interviews for research and reporting
02Creating searchable documentation or knowledge base articles from video tutorials
03Generating clean text summaries or study notes from long-form educational videos
