This project transforms GitHub repositories into engaging podcasts, targeting developers and tech enthusiasts who want to consume codebases audibly. It offers a novel way to digest project structures and documentation, making learning and exploration more accessible.

How It Works

The system leverages AI models for both content summarization (SSML generation) and text-to-speech (TTS). It parses repository structures and documentation, feeding this information into an AI model (OpenAI or Gemini Flash) to generate a script. This script is then converted into speech using the Azure Speech SDK, creating an audio podcast experience.

Quick Start & Requirements

• Install: Clone the repository, install dependencies (pnpm i), set up environment variables (.env with OpenAI/Azure keys), run backend (docker-compose up --build -d), start database (./start-database.sh), initialize schema (pnpm db:push), and run frontend (pnpm dev).
• Prerequisites: Node.js, pnpm, Docker, OpenAI API key, Azure AI Speech Key.
• Resources: Local development requires running backend, database, and frontend services.
• Links: GitHub Repo, Azure Speech Services

Highlighted Details

• Converts any GitHub repository into a podcast.
• Utilizes OpenAI/Gemini Flash for SSML and Azure Speech SDK for TTS.
• Offers API access (WIP) and self-hosting options.
• Frontend built with Next.js, Backend with FastAPI.

Maintenance & Community

• Open to contributions via Pull Requests.
• Mentions Gitingest and Gitdiagram for inspiration.

Licensing & Compatibility

• MIT License. Permissive for commercial use and closed-source linking.

Limitations & Caveats

The project is free to use with rate limits for hosted instances. Features like voice selection and API access are marked as "Work in Progress" (WIP). Users can bypass rate limits by self-hosting.