Surface B · The Voice Canvas

Tap to speak → the screen becomes the conversation

Everything else falls away. One focused surface cycles through listening → thinking → speaking, then waits for you. A single close icon returns to the full thread, where the exchange has been saved. Watch it run:

Listening
▍

One surface, three states

  • Listening — your voice, spiky reactive waveform in accent red
  • Thinking — the slow 1.9s breathe; the line goes still
  • Speaking — her words rise in serif as a calmer, slower waveform plays beneath
  • Then it waits — ready for your reply, no tapping required
Listen Think Speak

Decisions to settle: (1) Does every mic tap open this canvas, or keep a quick inline-listen for fast captures too? (2) Show Mallory’s words as text while she speaks (as shown — great for glance & accessibility), or audio-only and purer? (3) After she finishes, should it auto-listen for your reply (hands-free back-and-forth) or wait for a tap?

Tell me, and I’ll fold the voice canvas into the guide. This likely reshapes D too — the main chat becomes the place you go to read & type, with voice living here.