Skip to content
VocalFlow Docs

Context-Aware Transcription

Context features feed the recognition engine background information about what you are working on, so the terms you speak come out right. Context fixes mis-hearing; refinement fixes grammar and formatting — they complement each other.

Key points — the four context sources (all on by default, all can be turned off in Settings):

  • Window info: the front app’s name and window title are sent with the session, hinting at the domain you are talking about.
  • Browser URL: the current tab’s address (a sub-item of window info). Private/incognito windows are unconditionally skipped.
  • Editor text: the text around your cursor in the field you are dictating into. Password fields / secure input are unconditionally skipped.
  • Screenshot OCR: at the instant you press the hotkey, the front window is captured once and its on-screen text is read. Requires Screen Recording permission; silently skipped without it. Zero background screenshots — capture happens only at that instant.

The hard privacy rules, which no setting can override:

  • The setup wizard tells you explicitly which context features are on by default — informed defaults, not silent collection.
  • Secure input (password fields) and private browsing windows are always excluded from collection.
  • Screenshots happen only at the moment you press the hotkey, never in the background.
  • Any context source that fails is silently skipped — context never delays or interrupts your recording.