Audio Simulation

A classroom discussion simulation with AI-powered agents, featuring a Next.js frontend and a FastAPI backend.

Prerequisites

• Node.js (v18+)
• Python (v3.10+)
• npm (or yarn/pnpm/bun)
• pip

Setup

Backend (Python)

Install the required Python packages:

cd backend
pip install -r requirements.txt

Some evals have additional dependencies. For example, the stance detection coherence eval requires torch + transformers:

pip install -r backend/eval/stance_detection_coherence/requirements.txt

You will also need to set the following environment variables (e.g. in a .env file inside backend/):

OPENAI_API_KEY=your_openai_key
ELEVENLABS_API_KEY=your_elevenlabs_key

Note: If you have an OpenAI key saved in the env, you might need to got backend/agent_personas.py line 24 to override it.

Frontend (Next.js)

STEP1 :Install the npm dependencies:

cd frontend
npm install

This installs all packages listed in package.json, including: next, react, react-dom, @elevenlabs/elevenlabs-js, gtts, radix-ui, lucide-react, class-variance-authority, clsx, tailwind-merge, and dotenv.

STEP2: Replace the root in frontend/next.config.ts with your own project path and point it to the frontend

Starting the Program

From the project root run:

bash start.sh

Open http://localhost:3000 with your browser to see the result.