tl;dl is an AI-powered podcast summarization tool that converts Google Podcasts episodes into concise summaries. It downloads podcast episodes, transcribes them, generates summaries, and can even create audio versions of the summaries.
- 🎙️ Download podcasts from YouTube Music Podcasts
- 📝 Transcribe audio to text using OpenAI's Whisper
- 🤖 Generate summaries using local LLM through Ollama
- 🗣️ Convert summaries back to speech using TTS
- 🌐 User-friendly web interface with Streamlit
The project uses the following key technologies:
- Streamlit - Web interface
- Whisper - Speech-to-text transcription
- Ollama - Local LLM for summarization
- TTS - Text-to-speech synthesis
- Python 3.10.12 or higher
- Ollama installed and running locally
- At least 8GB RAM recommended
- GPU recommended but not required
- Clone the repository:
git clone https://github.com/makermotion/tldl.git
cd tldl- Create and activate virtual environment:
python3.10 -m venv env
source env/bin/activate # On Windows: .\env\Scripts\activate- Install the package:
pip install .- Start Ollama server with the required model:
ollama run <model> - Start the web interface:
streamlit run src/app.py- Open your browser and navigate to
http://localhost:8501 - Search a podcast by its name like you do in YouTube music.
- Processing time depends on episode length and system capabilities.
Contributions are welcome! Here are some ways you can help:
- Submit bug fixes or improvements via pull requests
- Help with documentation
- Suggest new features
Current development priorities:
- Command-line interface implementation
- Docker containerization
- Improved podcast source compatibility
- Performance optimizations
If you encounter any issues or have questions:
- Create an issue
- Check existing issues for solutions
Released under the MIT License. See LICENSE for details.