🌀 relife

What if you had chosen differently? — A multi-modal life simulation that reads your emotional state via EEG and generates branching alternate-life scenarios with AI-written narrative, generated imagery, and a matching soundtrack.

Built with the CAMEL AI agent framework. Each turn, the system presents 2–3 diverging choices (90% unexpected, 9.9% butterfly-effect, 0.1% ideal ending). Your EEG emotion feeds continuously into the music generation layer, keeping the soundtrack in sync with how you actually feel.

What it does

• Narrative engine — CAMEL multi-agent system generates branching life scenarios from your memories. 8–10 turns per session, with memory anchors stored across sessions
• Image generation — Each scene is illustrated in real time (Volcengine / Doubao seedream-3-0), dreamcore + pixel art aesthetic
• EEG music — Emotiv headset → emotion state → Google Lyria generates a matching ambient track per scene (optional; runs without headset using manual emotion input)
• Frontend — React interface with scene display, choice selection, and live image/music progress

Architecture

Emotiv Headset (optional)
        │
        ▼
brain_processor.py ──▶ emotion state
                              │
              ┌───────────────┼─────────────────┐
              ▼               ▼                 ▼
      story_agent.py    imgen_tool.py    musicgen_tool.py
      (CAMEL + GPT)   (Doubao API)     (Google Lyria)
              │               │                 │
              └───────────────┴─────────────────┘
                              │
                       React frontend

Stack

• Agent framework: CAMEL AI
• LLM: GPT-4o mini (memory management)
• Image generation: Volcengine / Doubao doubao-seedream-3-0-t2i
• Music generation: Google Lyria real-time
• EEG (optional): Emotiv Cortex SDK
• Frontend: React + Vite
• Backend: Flask

Quick start

# Backend
pip install -r camel_agent/requirements.txt
# set OPENAI_API_KEY, ARK_API_KEY, GOOGLE_API_KEY in .env
python camel_agent/main.py

# Frontend
cd frontend && npm install && npm run dev

EEG headset is optional — the system runs with manual emotion input if no device is connected.