A command-line tool that automatically transcribes and summarizes YouTube videos using yt-dlp, Whisper, and Fabric/LLaMA. This tool is particularly useful for videos without available transcripts or when you need quick summaries of video content.
- Downloads audio from YouTube videos
- Transcribes speech to text using Whisper
- Generates summaries using Fabric/LLaMA models
- Handles temporary file cleanup automatically
The following tools need to be installed:
- yt-dlp - For YouTube video downloading
- ffmpeg - For audio processing
- Fabric - For text summarization
- Whisper - For speech-to-text conversion
brew install yt-dlp ffmpeg
go install github.com/danielmiessler/fabric@latest
read whisper.cpp/README.md for how to download models
# Summarize a single video using llama3.2
python summarize_yt.py -y "https://www.youtube.com/watch?v=dQw
python summarize_yt.py --youtube-url "https://youtu.be/abc123" --model-name "gpt4:latest"
Argument | Short | Required | Description |
---|---|---|---|
--youtube-url |
-y |
Yes | The URL of the YouTube video to process |
--model-name |
-m |
Yes | The Fabric/LLaMA model to use for summarization |