Freeflow is a native Swift dictation app for Mac. Free, open-source, and entirely on-device — your voice never leaves your machine.
Every feature ships on-device. Nothing you say is anyone's data.
Hold your key and talk — your words land at the cursor, in any app on your Mac.
Speech recognition runs on the Neural Engine. No servers, no API keys, no telemetry.
Transcription at 145× real-time. Your sentence appears before you finish the next one.
History, stats and themes — stored locally. Open source under MIT, forever.
macOS asks for two permissions. The model downloads once. Then it's yours.
Drag Freeflow to Applications and launch it. Signed and notarized.
A one-minute wizard handles permissions and the ~465 MB model download.
Hold ⌃ Control, talk naturally, release. Your words land at the cursor.
History, stats and themes — without the clutter.
No cloud processing. No telemetry. No analytics. The transcription model runs entirely on your device — what you say is your business, not a training dataset.
