✦World Room
Ready to open

A live room for imaginary worlds

Say it aloud.
Watch a world answer.

Speak with a playful realtime companion that remembers your canon, follows interruptions, and helps uncover places, people, conflicts, and scene hooks.

Before the door opens

Set the first spark

Your browser handles microphone capture and audio playback over WebRTC. Your server only mints a short-lived client token; the permanent API key never enters the browser.

Live chamber

Voice session

Not connected
Live transcriptTranscription can arrive after speech begins

Your spoken story will gather here after the session opens.

WEBRTC

Speech in, speech out

The browser uses OpenAI's low-latency realtime transport instead of a stitched text and TTS loop.

SEMANTIC VAD

Natural turns and interruption

The room listens for meaning, lets you cut in, and resumes from the latest direction.

EPHEMERAL ACCESS

The permanent key stays server-side

Every browser connection starts with a fresh short-lived client secret from your own API route.