Dictate a message and hear spoken replies

Updated October 28, 2026

On this page

Use supported browser speech features with the assistant.

Before you start

Use a browser that supports the displayed speech controls and a working microphone. The assistant must be connected.

Open an assistant conversation and find Start voice input beside the composer.

Choose Start voice input and allow microphone access when prompted.
Dictate your message.
Dictation stops automatically when you pause. Choose Stop voice input if you want to end it sooner.
Review and correct the transcribed text, then send the message.
If available, choose Enable spoken replies to hear responses. Use Mute spoken replies to turn speech off.

Speech is transcribed into the message composer. Sending the message remains a separate action.

Dictate a request for a caption draft, correct the product name in the transcription, then send it.

Speech recognition and playback depend on browser support.
Spoken replies read each reply once it has finished, using a voice installed on your device in your browser's language when there is one. Sending a new message stops the reply being read. Very long replies are read only in part; read the rest on screen.
Browser microphone permission is required for dictation.
Voice input does not bypass action approval or the assistant’s access limits.

Voice is disabled: check browser support and the assistant connection.
Microphone access was denied: review the browser’s permission for this site.
Words are wrong: edit the transcription before sending.
Spoken replies are on but you hear nothing: check your device volume and that your browser has a voice installed for your language, then send another message.