Recording Modes and Hotkeys

Recording Modes

Hold-to-talk: click into your destination field, hold Option, speak, then release to finish. The mode is enabled by default and can be disabled in Settings → Recording. Recording starts the moment Option goes down, so opening words are not clipped; the overlay appears once Steno can tell you're holding Option rather than typing a shortcut.

Keyboard shortcuts that use Option, such as Option+Arrow, Option+Delete, or Cmd+Option+I, and quick taps of Option don't start a dictation. To tell them apart, Steno notices key presses and mouse clicks in any app while Hold Option to talk is on. It uses only the fact that a key or button was pressed, never which one.

Hands-free: press your configured function key to start, then press it again to finish. The default is F18. The main Dictate control also provides recording controls.

Stop ends the recording, transcribes it, and inserts the result. Cancel discards the current recording, including while Steno is transcribing; nothing is inserted or saved to History. Neither action removes text from an earlier completed dictation. Recordings stop automatically after one hour, with a countdown in the overlay during the last minute, and the audio is transcribed.

Default Hotkeys

Settings → Recording offers Disabled or F1–F20 for the hands-free key. Disabled turns the key off; Hold Option and the Dictate button keep working. F1–F12, including F3 and F4, require standard function-key mode in System Settings; F13–F20 require a keyboard or mapping that can emit that key. Choose a key that does not conflict with another shortcut, then select Save changes. Steno suppresses its selected hands-free key event while its monitor handles it.

Global shortcuts depend on macOS event access. Check Input Monitoring and Accessibility if they do not work outside Steno.

State Transitions

Recording proceeds to final transcription when you release Option or choose Stop. If the release of Option is missed, for example around a password field or a screen lock, the recording ends within about a second and is transcribed. A dictation shortcut pressed while the previous dictation is still finishing shows a brief notice in the overlay instead of starting a new recording. With microphone access turned off, the dictation shortcuts start nothing and say that microphone access is off. If the microphone stops during a recording, for example when a USB or Bluetooth microphone is disconnected, the recording ends at once and what was said before it stopped is still transcribed. Releasing Option does not stop an independently started hands-free recording, and switching modes does not create overlapping sessions.

The optional live transcript is a local preview. Words may appear late or change during recording. Even a correct-looking preview can differ from the final text. Check the completed text and History outcome.

Nearby Text and Spoken Commands

Use nearby text for continuation is opt-in under Settings → Recording. It reads a limited range around the selection to adjust spacing and conservative capitalization. Automatic continuation currently supports English in supported fields in Apple Mail, Notes, and TextEdit. Secure or unsupported fields and an unconfirmed original editor target are excluded. Nearby text is temporary and is never added to recognition prompts, History, or Insights.

To lowercase the first cased character after cleanup, begin with lowercase followed by whitespace and your phrase: lowercase Hello there becomes hello there. This command works without nearby-text continuation. Matching ignores case and leading whitespace. lowercase alone, please lowercase Hello, and lowercase, hello stay literal. Quotes inside a matching payload do not prevent the command.

To retain the word, say literal lowercase followed by whitespace and nonempty text: literal lowercase Hello keeps lowercase Hello without applying the lowercase command.

Verify a Recording

Try a short sentence in your regular editor. Check opening words, Stop, Cancel, and clipboard recovery separately. Quiet speech after a long silence can lose opening words during voice detection, and Small can mishear words. Check the final text even if the preview looked correct.