What Voqalize is
The brain, the session, and the two halves of your app. Start here →
Voqalize runs the voice stack — WebRTC, voice activity detection, speech-to-text, text-to-speech, interruption handling, recording. You write the brain: an agent that receives the caller’s words and drives two channels back — speech, and actions that drive the screen the caller is looking at. Your code never touches audio.
What Voqalize is
The brain, the session, and the two halves of your app. Start here →
Put a voice in the browser
One HTTP request, then stock pipecat. The handshake →
Test a brain without a microphone
Drive your real brain over the real wire in text mode. Testing a brain →
Where the brain runs
Your own inbound server, or a relay when you can’t accept inbound. Deploy →
A brain is a single WebSocket URL. Voqalize dials {brain_url}?session_id={session_id},
one connection per session, opened when a call starts and torn down when it ends.
Where that URL points is up to you — your own inbound server, or a Cortex relay
for brains that can’t accept inbound connections. Same brain code either way;
you only choose who dials whom. See Where the brain runs.