This project is an interactive playground that demonstrates the capabilities of OpenAI's Realtime API, allowing users to experiment with the API directly in their browser. It's built on top of LiveKit Agents.
See it in action at realtime-playground.livekit.io
This directory contains the agent implementation. There are two versions, one in Python and one in TypeScript.
playground_agent.py
- Built on the LiveKit Python Agents frameworkplayground_agent.ts
- Built on the LiveKit Node.js Agents frameworkThis directory houses the web frontend, built with Next.js.
/agent
directorycp .env.sample .env.local
.env.local
in a text editor and enter your LiveKit credentialspython -m venv .venv
source .venv/bin/activate
.venv\Scripts\activate
source .env.local
set -a; . .env.local; set +a
pip install -r requirements.txt
python playground_agent.py dev
pnpm install
source .env.local
set -a; . .env.local; set +a
pnpm dev
/web
directorycp .env.sample .env.local
.env.local
in a text editor and enter your LiveKit credentials:pnpm install
pnpm dev
The agent can be deployed in a variety of ways: Deployment & Scaling Guide
The web frontend can be deployed using your preferred Next.js hosting solution, such as Vercel.
Ensure the following:
For more information or support, please refer to LiveKit docs.
Apache 2.0