Inside Inside Out

A real-time voice AI console where five emotions talk, debate, and interrupt each other

What is this?

Inside Inside Out is a voice-first AI console built for the AI Accelerate: Unlocking New Frontiers challenge (Google Cloud + ElevenLabs).

Inspired by Pixar's Inside Out, this project lets you talk to five AI personalities (Joy, Sadness, Anger, Fear, and Disgust) who don't just respond to you. They debate with each other, interrupt, and react in real-time, creating genuinely dynamic conversations.

How it works

We use a "think slow, act fast" approach:

The Brain → Gemini 2.5 Flash Lite analyzes your input and decides which emotion should respond and what they should say.

The Voice → ElevenLabs Turbo v2.5 streams the speech back with ultra-low latency, so conversations feel natural and immediate.

The Interface → A React app with real-time audio visualization and smooth WebSocket streaming.

Tech Stack

AI & Intelligence

• Google Gemini 2.5 Flash Lite (via Vertex AI): Handles emotional reasoning and dialogue generation
• Fast enough for real-time responses, smart enough for personality consistency

Voice Synthesis

• ElevenLabs Turbo v2.5: Converts text to speech with minimal latency
• Each emotion has a distinct voice with expressive capabilities (sighs, shouts, whispers)

Frontend

• React + Vite: Fast dev experience and smooth UI
• Framer Motion: Polished animations
• WebSocket streaming: Real-time audio playback

Backend

• FastAPI: Handles routing and orchestration
• Streaming architecture: Audio chunks flow continuously from ElevenLabs to the browser

Features

War Room (Call Mode)

Have a continuous voice conversation with all five emotions. They'll debate your life choices, argue with each other, and interrupt when they feel strongly about something.

Fun Mode

Give the emotions a scenario (like "It's Monday tomorrow") and watch them perform a scripted comedy skit. The Brain writes the dialogue on the fly, and all five emotions act it out.

Judge Mode

For demos and hackathons. Just paste your API keys (Gemini + ElevenLabs) in the UI, and the system reconfigures itself instantly - no backend setup needed.

Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• Gemini API Key (Google Cloud)
• ElevenLabs API Key

Quick Start

1. Clone and navigate

   git clone https://github.com/PratikCreates/Inside-Inside-Out.git
   cd Inside-Inside-Out

2. Add your API keys

   # Copy the example file
   cp .env.example .env
   
   # Edit .env and add:
   # GEMINI_API_KEY=your_key_here
   # ELEVENLABS_API_KEY=your_key_here

3. Install dependencies

   cd backend && pip install -r requirements.txt && cd ..
   cd frontend && npm install && cd ..

4. Launch the app

   Windows (Recommended):

   .\run_app.bat

   Or manually (any platform):

   # Terminal 1 - Backend
   cd backend && uvicorn main:app --reload
   
   # Terminal 2 - Frontend
   cd frontend && npm run dev

5. Open http://localhost:5173 🎉

Troubleshooting

PowerShell: "run_app.bat is not recognized"

PowerShell requires .\ prefix for local scripts:

.\run_app.bat

Port already in use

Kill the process using the port:

npx kill-port 8000 5173

Missing dependencies

Make sure you ran the install commands:

cd backend && pip install -r requirements.txt
cd frontend && npm install

Contributing

Pull requests are welcome. Some ideas: add new emotions, optimize audio streaming, or improve the orchestration logic.

---

Built for the AI Accelerate hackathon