Voice Mode
Talk to Alvin instead of typing, and have it talk back.
Voice mode turns the chat into a conversation you can have out loud. You speak, Alvin answers in a natural voice, and everything else — routing, canvases, agents — works exactly as it does in text.
Using it
- Open a chat
- Click the microphone
- Talk
Speech is transcribed as you go, and replies are spoken back. You can switch to typing mid-conversation and back again — it's the same conversation either way.
What it's good for
- Thinking out loud while your hands are busy
- Working through a problem conversationally, where typing slows you down
- Accessibility — no typing required
- Longer dictation, where speaking is faster than writing
Availability
Real-time voice conversation is available from Standard. Spoken output and transcription are available on every plan. See Limits & quotas.
Notes
Voice consumes model usage the same way text does — the transcription and the speech synthesis are both metered. A long spoken session costs more than the equivalent typed one.
For a quiet room, expect the transcription to be close to exact. In noisy environments it's worth checking what was heard before acting on the answer.
Related
- AI Chat — the chat interface
- Turbo mode — several models at once
- Plans — what each plan includes