The Andrew Exchange

A web application that discovers trending political topics, analyzes debates from both liberal (Andrew Carnegie) and conservative (Andrew Mellon) perspectives, generates debate scripts, and creates AI-powered audio with distinct voices for each side.

Tech Stack

Frontend:

• React
• TypeScript
• Tailwind CSS

Backend:

• FastAPI (Python)
• OpenRouter API (Perplexity Sonar Pro Search for debate analysis, GPT-4o for script generation)
• Eleven Labs API (AI Text-to-Speech for audio generation)

Quick Start Guide

1. Open New Terminal (Terminal 1)

2. Backend Command:

python3 -m uvicorn app:app --reload

3. Open New Terminal (Terminal 2)

4. Frontend Command:

npm start

5. Use the App

1. Open in your browser
2. See trending political topics displayed at the top
3. Either:
   • Click a trending topic to select it
   • Or enter your own custom topic
4. Click "Analyze" - this will:
   • Generate debate arguments (liberal vs conservative)
   • Create a conversational script between Carnegie (liberal) and Mellon (conservative)
   • Generate audio files with distinct AI voices for each speaker
5. Listen to the debate by clicking "Play Debate"
6. Watch the speaker portraits animate as each side speaks

How It Works

The app runs a complete AI pipeline when you analyze a topic:

1. Trending Topics → News API fetches recent political articles → AI categorizes them into trending topics
2. Topic Analysis → Perplexity Sonar Pro Search analyzes both sides of the debate with recent sources
3. Script Generation → GPT-4o creates a natural conversation script between Carnegie (liberal) and Mellon (conservative)
4. Audio Generation → Eleven Labs converts each speaker's lines into audio with distinct voices
5. Playback → Frontend plays the audio files in sequence with visual indicators

Project Structure

NovaHacks2025/
├── app.py                      # FastAPI backend API (main server)
├── search.py                   # Debate analysis logic
├── script_generator.py         # AI script generation
├── videogenerator.py           # Audio generation with Eleven Labs
├── requirements.txt            # Python dependencies
├── .env                        # API keys (OPENROUTER_API_KEY, ELEVENLABS_API_KEY, NEWS_API_KEY)
├── results/                    # Saved debate JSON files
├── audio_output/               # Generated audio files for debates
├── politics_news_scraper/      # News scraping & categorization
│   ├── news_scraper.py         # Fetches news from News API
│   ├── categorizer.py          # AI topic categorization
│   └── config.py               # Configuration
└── frontend/                   # React frontend
    ├── src/
    │   ├── App.tsx             # Main component with audio player
    │   └── index.css           # Tailwind CSS
    ├── public/
    │   ├── carnegie.jpg        # Portrait of Andrew Carnegie (liberal)
    │   └── mellon.jpg          # Portrait of Andrew Mellon (conservative)
    ├── tailwind.config.js      # Tailwind config
    └── package.json            # Node dependencies