Skip to main content
Get an API key from the dashboard, paste it below, and select Start recording. Allow microphone access and speak. Select Stop and transcribe to receive the final transcript. Auto-finalize after pauses confirms each utterance after 500 milliseconds of silence and keeps listening. Turn it off to receive one final transcript when you stop. Choose Balanced for faster feedback or High accuracy for more context. Language detection is automatic. The demo uses the same ASR WebSocket API as your application and charges your workspace’s normal speech-to-text credits. The two-minute recording cap applies to this demo, not the API.
Closing or reloading the page stops the recording and clears the transcript and API key from the demo. Copy any text you want to keep before leaving. For application integration, keep your API key on your server.

​
Inspect the API messages

The WebSocket reference also has a developer playground. Enter your API key under Security Schemes, select Connect, and wait for session.started.
  • Keep alive resets the idle timer. It has no response.
  • Commit a segment finalizes audio sent so far and keeps the session open.
  • End input returns the remaining transcript and closes the connection.
These control messages are prefilled. To send microphone audio, use the demo above. See the streaming guide for the full protocol and quickstart for runnable client code.