GPT-Twin

README excerpt

# Digital Twin API

A complete workflow system that generates personalized hot take responses from various personas using AI-powered audio and video processing.

## Features

🎬 **Audio/Video Processing**: Convert speech to text using OpenAI Whisper
💬 **AI Hot Takes**: Generate responses using GPT-5 with customizable personas
🎵 **Voice Synthesis**: Convert text to speech using ElevenLabs or HeyGen
🎥 **Avatar Videos**: Create talking avatar videos using HeyGen
👥 **Multi-Persona Support**: Switch between different character personalities
🔧 **Modular Workflow**: Process each step independently or as complete workflows

## Available Personas

- **Chad Goldstein**: Flamboyant venture capitalist with tech-bro energy
- **Sarah Guo**: Analytical VC from Conviction with sharp insights
- **Sarah Chen**: Tech journalist with industry expertise
- **Custom Personas**: Create your own with unique prompts and voices

## Workflow

```
Audio/Video Input → Transcript → Hot Take → Voice → Avatar Video
```

**Or process steps independently:**
- `POST /generate-text` - GPT text generation only
- `POST /generate-audio` - ElevenLabs audio generation from text
- `POST /generate-video` - HeyGen video generation from text or audio

## Installation

1. **Clone and setup**:
```bash
git clone <repository>
cd digital-twin
pip install -r requirements.txt
```

2. **Configure environment variables**:
Create a `.env` file with your API keys:
```bash
# API Keys (Required)
OPENAI_API_KEY=your_openai_api_key_here