Voice lets you speak to the workspace instead of typing, and have the AI read its responses back to you. All voice interactions flow through the gateway, so your team’s existing rate limits, audit logging, and permissions apply.

​
Voice interaction modes

Push-to-talk

Click the mic button in the chat input to start recording. Release to transcribe your speech. The transcript is inserted into the input field so you can review or edit it before sending.

Read aloud

Every assistant message has a play button. Click it to have the response read aloud using the configured voice. Click stop to end playback early.

Voice loop

Continuous hands-free mode: record → transcribe → send → hear the reply, then repeat automatically. Useful for eyes-free use cases or accessibility.

​
Getting started with voice

1

Check your permissions

Voice must be enabled for your team. If you don’t see the microphone icon in the chat toolbar, contact your admin to enable the voice feature permission for your team.
2

Allow microphone access

The first time you use push-to-talk, your browser will ask for microphone permission. Click Allow.
3

Use push-to-talk

Click the mic icon in the chat input. Speak your message, then click the icon again (or release, depending on your browser) to stop recording. Your transcribed text appears in the input field. Edit if needed, then send.
4

Read responses aloud

Click the speaker icon on any assistant message to hear it read aloud. Use this with push-to-talk for a fully voice-driven conversation.

​
Admin configuration

Configure voice in Admin UI → Settings → Voice: 
voice:
  enabled: true
  stt_model: whisper-1      # speech-to-text model
  tts_model: tts-1          # text-to-speech model
  tts_voice: alloy          # default voice character
Available voice characters: alloy, echo, fable, onyx, nova, shimmer.

​
Access control

The voice feature permission must be enabled per-team in Admin UI → Teams → Permissions. Users on teams without the permission will not see voice controls in the chat UI.