Help·Chat and models 2 min
Voice conversation: talk instead of type
Live conversation like on the phone — with voice of your choice, mute and web search.
The wave icon in the input field starts a conversation: you speak, the AI replies by voice, everything is transcribed and appears in the chat.
- 1
Getting started
In the chat, tap the wave icon on the right in the input field, allow microphone access. The orb shows the state: Listening, Thinking, Speaking.
- 2
Speaking and interrupting
Just start talking, after a short pause the AI replies. You can interrupt it at any time. If you ask about current events like weather or news, it says ’I’ll look that up’ — for weather it fetches real forecast data (progression throughout the day, rain, wind), for other things it searches the web.
- 3
Files and images by voice
Say ’Create an Excel with…’, ’Write that as a Word document’, ’Turn that into a PDF’, ’Generate an image of…’ or ’Create a presentation about…’. The assistant says ’I’ll have that created’, the studio works in the background (usually 10 to 60 seconds), and the file or image appears in the chat — you can keep talking in the meantime. It also passes tricky questions, calculations, analyses and translations to the powerful text model: you hear the short answer, the detailed one is in the chat.
- 4
Mute
The microphone button (or M key) mutes your microphone — handy when there’s conversation in the room. The red button ends the conversation.
- 5
Choose voice
Settings → Account → ’Voice in conversation’: ten voices with descriptions, applies from the next conversation onwards.
Tips
- Headphones give the most reliable recognition, because there’s no speaker echo.
- Your memory (e.g. location) is also known to the voice assistant — ’What’s the weather like?’ is enough.
Cost: Spoken minutes cost more than text, roughly 15 to 30 cents per minute of conversation; shown under usage.