OpenAI Realtime API Demos

This page shows demos of the OpenAI Realtime API, which makes it easy to build realtime voice and text AI applications. The demos are written in Next.js with TypeScript, integrated with Google Vertex AI UI. Click the GitHub link below to view the source code. Patches and issues welcome!

Note that the demos require an OpenAI API key, which you can get here.

🤖 Voice Agent DemoLets you chat with a simple voice agent.
📓 Voice to Text Agent DemoLets you chat with an agent that listens to voice and replies with text.
📝 Transcription DemoTranscribes mic or audio file input.
🔁 Loopback DemoEchoes back what you say after applying noise reduction.
🎨 Image Generation DemoDemonstrates how to generate images from a voice conversation.
🧑‍💻 Vibe Coding DemoDemonstrates how to delegate to a reasoning model for code generation.