Foni is voice typing for Mac. Clean, corrected text lands at your cursor about a second after you let go. Nothing leaves your machine.
Press and hold Right ⌥ in any app. Tap it instead to toggle. Works over fullscreen windows.
Ums, false starts, “scratch that”. Say “new paragraph”. Speak like a person, not a printer.
Clean text pastes at your cursor. Your clipboard is put back the way it was.
Both models stay loaded and warm the whole time. The correction model even remembers your context between dictations, so it never re-reads it.
Measured on an M4 Pro. Your Mac may swim at a slightly different speed.
Transcription and correction both run on your Mac. Unplug the internet and Foni keeps working exactly the same.
After each paste, Foni watches what you edit. Correct the same mishearing twice and it proposes a fix. One click and that name, product or acronym is caught for good.
Everything it learns lives in a markdown file. Read it, edit it, put it in git.
# ~/.config/foni/mishearings.md ## Auto-learned - "sara chan", "sarah chen" -> Sara Chen - "fony fish", "phony" -> Foni - "post gress", "postgres sequel" -> PostgreSQL