Hold a key anywhere — editor, browser, terminal, chat — and talk. Release, and the words are at your caret before your hands are back on the keys. No window to open, no app to switch to.
Get notified at launchFlowSpeak types by borrowing your clipboard for a blink and putting it back exactly as it was — text, images, files, formatting, all restored. Its little status pill never takes focus, so your editor never loses the caret. And if a paste can't land, the transcript is still on your clipboard and in your history — words are never lost.
Audio streams as you speak, so the finished text is essentially ready the moment you release the key. You watch the words assemble live in the pill. Accidental taps cost nothing — under a beat, nothing fires — and the mic buffer means even your first syllable isn't clipped.
Boost the names and jargon it should expect, and add replacement rules — say the long thing, type the short thing. Rules apply to the very next dictation, no restart. Filler words are stripped on the way through: "this is, um, a test" comes out clean.
Your last 200 dictations live in a plain local file on your machine, listed newest-first in the History window. Click to copy; anything that failed to paste is badged so you can rescue it. Toggle history off, or clear it, any time. Nothing is uploaded.
If transcription stalls, FlowSpeak retries with the full recorded audio. If it still can't get through, it saves the recording locally and tells you where. The words you said exist somewhere, always.
No seats, no tiers, no per-word math — pay for the tool, use the tool. macOS first; Windows is on the bench being fitted.
FlowSpeak is in development for macOS and Windows. Leave your email and we'll tell you the moment it ships.