Skip to main content
Silent start lets you begin a voice session without an opening greeting from Athena. Use it when your application should let the participant speak first. The session stays quiet until someone speaks, and it does not add an opening greeting to the transcript. You configure this behavior when you create the session.

​
Configure silent start with the React SDK

  1. In your application, use useAthenaVoiceSession from @athenaintel/react/voice.
  2. For a session the hook creates itself, set silent_start to true inside the session configuration.
  3. Start the session and allow microphone access.
  4. Speak to begin the conversation. Check that the first exchange starts with your words rather than an agent greeting.
The hook can create sessions itself when your app runs at a Computer-hosted application URL. For an app running elsewhere, supply a createSession callback that obtains session credentials through your server.

​
Configure a session created by your server

  1. Create the voice session on your server using the authenticated voice-session API.
  2. Include silent_start: true in the session request body.
  3. Return the session credentials to your application through createSession.
  4. Connect and speak to begin the conversation.
When you provide createSession, the hook ignores its session configuration. Set silent start in your server’s request instead. Keep API keys on your server.

​
Good to know

Silent start defaults to false. You must opt in when creating a session. It changes the opening behavior; it does not mute the microphone or prevent Athena from responding after you speak. A resumed silent-start session also skips the greeting when you rejoin. Voice must be available in your deployment. Silent start does not apply to outbound phone calls. Those calls clear this option so Athena can speak first when the recipient answers. Do not use a silent-start test to infer the opening behavior of an outbound call.