Mr.🆖 PodcastAI 🎙️🎧

Transform documents, images, and websites into engaging podcast episodes using AI

Convert long-form content into natural podcast dialogues that capture attention and make information more accessible for auditory learning on the go.

Features

• 📁 Versatile Input Support: Upload documents (PDF, DOCX, TXT), images with OCR (JPG, JPEG, PNG), paste text directly, or convert websites via URL
• 🤖 AI-Powered Dialogue Generation: Uses OpenAI's GPT-4.1-mini to create natural, engaging podcast conversations from your content
• 🎵 Professional Audio: Leverages OpenAI's text-to-speech models for high-quality, lifelike voices through Mr.🆖 AI Hub routing
• 🌐 Multi-Language Support: Generate podcasts in English, Chinese (Traditional), or Cantonese with optimized voice synthesis
• 💰 Cost Transparency: Real-time TTS cost calculation and tracking (English, Chinese, Cantonese)
• 🖥️ User-Friendly Interface: Gradio-based web interface for easy interaction
• 💾 Smart History Management: Browse and reload previous podcasts stored in your browser (IndexedDB + localStorage)
• 🔧 Resilient Processing: Retry mechanisms and error handling for reliable conversion
• ⚡ FastAPI Backend: Robust server architecture with deployment-ready setup

Demo Examples

The project includes sample inputs:

• PDF documents (e.g., "Intangible cultural heritage item.pdf")
• Images with text (e.g., "JUPAS Guide.jpg")
• URL extraction from web pages

Installation

Prerequisites

• Python >= 3.12
• uv package manager (recommended)

Quick Setup

1. Clone the repository:

   git clone https://github.com/bentwnghk/mr5-podcast-ai.git
   cd mr5-podcast-ai

2. Install dependencies:

   uv sync

3. Environment Setup:

   • Get an API key from Mr.🆖 AI Hub
   • Set environment variables:

     export OPENAI_API_KEY="your-api-key-here"
     export OPENAI_BASE_URL="https://api.mr5ai.com/v1"  # Mr.🆖 AI Hub endpoint

   • Optional: Configure Sentry for error monitoring:

     export SENTRY_DSN="your-sentry-dsn"

Usage

Quick Start

1. Launch the Application:

   uv run python main.py

   The Gradio interface will open in your browser at http://localhost:8000

2. Generate Podcasts:

   • Upload Files: Select PDF, DOCX, TXT, or image files
   • Paste Text: Directly input text content
   • From URL: Convert web pages to podcasts
   • Choose language: English, Chinese, or Cantonese
   • Enter your API key (auto-saved to browser)
   • Click "Generate Podcast"

3. View Results:

   • Listen to the generated MP3 podcast
   • Review the dialogue transcript
   • Check TTS costs
   • Access previous podcasts in the History panel

Input Methods

1. File Upload

• PDF: Digital documents, scans, reports
• DOCX: Word documents
• TXT: Plain text files
• Images: JPG/PNG with text extraction via OpenAI Vision API

2. Text Input

• Paste any text content
• Supports up to ~8000 token dialogues

3. URL Processing

• Convert web articles to podcasts
• Automatic content extraction with fallbacks
• Supports major news sites and blogs

Language Options

• English: Standard OpenAI TTS models
• Chinese (繁體): Traditional Chinese with optimized output
• Cantonese: Specialized voice support through Mr.🆖 AI Hub

Architecture

• Frontend: Gradio web interface
• Backend: FastAPI server with async processing
• Storage: Temporary file management with auto-cleanup
• AI Services: OpenAI GPT-4.1-mini + TTS via Mr.🆖 AI Hub
• Database: Browser-based history (IndexedDB/localStorage)
• Deployment: Ready for Docker/Uvicorn

Cost Estimation

TTS costs vary by language:

• English: ~$0.15 per 1M characters
• Chinese: ~$0.30 per 1M characters (x2 multiplier)
• Cantonese: ~$0.75 per 1M characters (x8 multiplier)

Project Structure

.
├── main.py              # Application entry point
├── description.md        # UI descriptions
├── head.html             # Custom HTML/JS for browser features
├── static/               # Web assets (logo, icon)
├── examples/             # Sample files for testing
├── pyproject.toml        # Python dependencies
├── uv.lock              # Dependency lock file
├── Dockerfile           # Container configuration
├── docker-compose.yml   # Docker composition
├── LICENSE              # Apache 2.0 License
└── README.md            # This file

Configuration

Environment Variables

Variable	Description	Required
OPENAI_API_KEY	Mr.🆖 AI Hub API key	Yes
OPENAI_BASE_URL	Mr.🆖 AI Hub endpoint URL	Yes
SENTRY_DSN	Sentry monitoring DSN	No

Custom API Endpoints

The application is designed to work with Mr.🆖 AI Hub compatible endpoints. Set OPENAI_BASE_URL to:

• Production: https://api.mr5ai.com/v1
• Local: http://localhost:3000/v1 (if running locally)

Troubleshooting

Common Issues

• API Key Issues: Ensure your Mr.🆖 AI Hub key is valid and has sufficient credits
• File Upload Errors: Check file size limits and supported formats
• URL Processing: Some websites block scraping - try different sources
• TTS Failures: Request timeouts - the app has retry mechanisms

Debug Mode

Set python -c "import logging; logging.basicConfig(level=logging.DEBUG)" before launching for detailed logs.

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature-name
3. Make your changes with tests
4. Submit a pull request

License

Licensed under the Apache License 2.0. See LICENSE for details.

Support

• Create issues on GitHub for bugs/feature requests
• Check the examples directory for sample inputs
• Review browser console for detailed error messages

---

Transform your content into podcasts that engage and inform 🎙️