An AI-powered web app that summarizes any YouTube video in seconds. Paste a link, click summarize, and get a clean summary — no need to watch the full video.
Built with Python, Streamlit, and Groq's LLaMA 3.3 70B model.
Paste any YouTube URL → Get an instant AI summary
Supports both URL formats:
https://www.youtube.com/watch?v=VIDEO_IDhttps://youtu.be/VIDEO_ID
- 🔗 Supports both standard and shortened YouTube URLs
- 🧠 Uses LLaMA 3.3 70B via Groq for fast, accurate summaries
- 📄 Handles long videos with transcript chunking
- ⚡ Clean and simple Streamlit UI
- 🌐 English video support
| Tool | Purpose |
|---|---|
| Python | Core language |
| Streamlit | Web UI |
| Groq API | LLM inference (LLaMA 3.3 70B) |
| youtube-transcript-api | Fetching video transcripts |
| python-dotenv | Managing API keys |
1. Clone the repository
git clone https://github.com/Darshan-2118/yt-summarizer
cd yt-summarizer2. Create and activate a virtual environment
python -m venv .venv
# Windows
.venv\Scripts\activate
# Mac/Linux
source .venv/bin/activate3. Install dependencies
pip install -r requirements.txt4. Set up your API key
Create a .env file in the root folder:
GROQ_API_KEY=your_groq_api_key_here
Get your free API key at console.groq.com
5. Run the app
streamlit run app.py| Variable | Description |
|---|---|
GROQ_API_KEY |
Your Groq API key (free at console.groq.com) |
yt-summarizer/
├── app.py # Main application
├── requirements.txt # Dependencies
├── .env # API keys (not pushed to GitHub)
├── .gitignore # Ignores .env and .venv
└── README.md # You are here
- Only supports videos with English transcripts
- Best results on videos under 20 minutes
- Requires an active internet connection
MIT License — feel free to use and modify this project.