Hold Option and talk - a global overlay captures speech into the Code composer on Mac. On iPhone, the Continuum Voice keyboard works in any app’s text field. Local models when you want offline; hands-free when your hands are full.
Lisbon · fable-5 · xhighworktree · lb-1Hold Option. Speak. Release. Text lands in the prompt.
A global Option-hold (configurable) brings up a Continuum voice overlay - waveform, partial transcript, silence countdown - without leaving the app you’re in. Release to stop, or let silence auto-stop finish the take.
There’s also an in-composer mic when you’re already in Code. Same engine, same models, same silence settings. Web and desktop composers reach the same stack through the voice bridge when a Mac is paired.
Pick a local engine: Apple Speech, WhisperKit, or Parakeet. Download the model once; run offline after that. No cloud STT hop for the take you don’t want to leave the laptop.
Parakeet is the fast path for long dictations. WhisperKit covers multilingual. Apple Speech is zero-download on supported system locales. Switch in Settings → Voice - the overlay and the composer mic both honor the choice.
The Continuum Voice keyboard is not locked to Continuum. Enable it once, switch to it from any app - Messages, Mail, Notes, Safari - and dictate. Text inserts at the cursor; if the host field won’t accept it, Continuum falls back to the clipboard.
While you record, Dynamic Island and a Live Activity show the take so you can see status without opening the app. Watch steers sessions; it doesn’t host speech recognition.
When your hands are on a build or a train pole, Continuum still takes the take.
Map Continuum Toggle Voice Control to the Action Button or a Control Center tile. One press starts or stops a take - no keyboard, no hotkey chord.
Double-tap the back of the iPhone to toggle the same voice control. Same insert path - keyboard field when focused, clipboard fallback otherwise.
Pick 1.5s / 2.5s / 4s of silence to end the take automatically. Hold-to-talk and hands-free starts both honor it - so you never have to hunt for Stop.
Short answers here, long answers in the docs.
Local engines (Apple Speech, WhisperKit, Parakeet) run on-device after the model is downloaded. That is the point of the local stack. If you choose a cloud path in a future build, it will be explicit - never a silent hop.
Yes. The global hold key is configurable in Settings → Voice. The default is Option-hold so it stays out of the way of normal typing and common app shortcuts.
Yes. Continuum Voice is a system keyboard - any text field in any app. Enable it under iOS Settings → General → Keyboard, then switch to it from the globe key when you need a take.
No. Watch is for session steering - approve, interrupt, glance. STT is Mac + iPhone only. That keeps the speech stack honest about where the mic and the models actually live.
Download Continuum, grant the mic, hold Option.
local models · iOS keyboard · hands-free starts