Talk to your Box (voice)

DocsUse your Box

Talk to your Box (voice)

Speak instead of typing, and hear replies read aloud in your Box's own voice. Voice works at your Box's secure Tailscale address.

Talk to your Box instead of typing, and hear its replies read aloud. Your Box does the listening and the speaking itself: it turns speech into text with Whisper and speaks with Kokoro, two open models that run on your Box.

Open your Box at its secure address

Browsers only let a page use your microphone over a secure (https://) connection. For your Box, that is its Tailscale address.

  1. Set up Remote access and turn on secure access. See Reach your Box from anywhere.
  2. Open your Box at the https:// address shown next to Address on the Remote access page, and sign in. Your browser shows a padlock.
The chat page at http://box.local with the message "Voice mode is only available on Tailscale." at the top right, and its Configure button circled

Dictate a message

  1. Click the microphone (Dictate) in the message box.
  2. The first time, your browser asks to use the microphone. Click Allow.
  3. Speak, then click the microphone again (Stop recording).
  4. Your words appear in the message box. Change them if you like, then press Enter to send.
The message box with the microphone Dictate button circled

Talk hands-free

Hands-free mode is a spoken conversation. Your Box listens, sends what you said, reads its reply aloud, and listens again.

  1. Click the waveform button (Talk (on this Box)), just right of the microphone.
The message box with the waveform Talk (on this Box) button circled, just right of the microphone
  1. Speak. When you pause, your Box sends your words.
  2. Your Box answers out loud, then listens for your next question.
  3. To stop, click the waveform button again.

The line above the message box shows what your Box is doing: Listening (on this Box), Transcribing on this Box, Sending, Thinking, then Speaking.

The message box during hands-free mode, with the status line "Listening (on this Box)" circled

Good to know

  • Voice understands English only.
  • Your Box reads replies aloud up to about 4,000 characters. A longer reply shows Too long to read aloud.
  • We have checked voice in Chrome and Safari on a Mac, and in Safari on an iPhone.

If something goes wrong

  • "Voice mode is only available on Tailscale." You opened your Box at http://. Click Configure, then open the https:// address on the Remote access page.
  • "Microphone access denied" or "Microphone unavailable". Your browser has blocked the microphone for your Box. Open the browser's settings for this site (in Chrome, the icon left of the address; in Safari, Safari → Settings for This Website), set Microphone to Allow, and reload the page.
  • "Box voice is still getting ready. Try again in a few minutes." Right after an update, your Box's voice can take a few minutes to start.
  • "Transcribing failed on this Box. Try again." Click the button and speak again.
  • You hear nothing. Check your device's volume.

Still stuck? See Get help. More about the chat screen: Chat with your Box.

Verified on Box software 0.2.11 · 2026-10-28