This project transforms academic research papers into engaging three-person podcast discussions, making complex information accessible to podcast enthusiasts and those who prefer auditory learning. It offers a unique way to digest research by simulating a natural, conversational flow.

How It Works

The tool employs a multi-stage LLM process: a Planning Chain structures the discussion around the paper's sections, a Discussion Chain uses RAG to generate dialogue based on retrieved content, and an Enhancement Chain refines the script for flow and clarity. Finally, the OpenAI API converts the script into audio with distinct personas (Host, Learner, Expert) for a realistic listening experience.

Quick Start & Requirements

• Install via git clone https://github.com/Azzedde/paper_to_podcast.git and cd paper_to_podcast.
• Requires a valid OpenAI API key in a .env file.
• Run with python paper_to_podcast.py path/to/your/research_paper.pdf.
• Place PDF research papers in the project directory.
• Sample podcasts are available in ./sample_podcasts.

Highlighted Details

• Simulates a three-person discussion (Host, Learner, Expert) for enhanced comprehension.
• Utilizes a RAG model to ensure script accuracy against source content.
• Cost-effective: ~$0.16 for a 9-minute podcast from a 19-page paper.

Maintenance & Community

• Open issue for runtime optimization.
• Future plans include local LLM and TTS integration (Ollama, open-source TTS).
• Contributions are welcomed via GitHub issues.

Licensing & Compatibility

• License not specified in the README.
• Compatibility for commercial use or closed-source linking is undetermined.

Limitations & Caveats

The current generation process is time-consuming, with optimization planned. The project relies on the OpenAI API, incurring costs and external dependencies.