🖥️
Runs entirely on your Mac
A stack of local recognizers turns speech into text on-device. No cloud, no API keys. Pull the Wi-Fi and it still works.
Local voice input for Mac
Hold a key.
Say the thing.
It types.
A one-click voice keyboard for your Mac. It listens, tidies up what you said, and drops the text right where your cursor is.
- ◆ 100% on-device
- ◆ No account
- ◆ No word cap
- ◆ Open source
Hold ⌥
Why it feels different
An interface between thought and software.
Not another transcript box. Every part is built to earn your trust with what you actually said.
🎯
Types into the right place
It leases the field under your cursor and checks it is still there before pasting once. If focus moved, the text parks safely in your Outbox instead of the wrong box.
⚡
Fast enough to forget
Models stay warm and long sentences are recognized while you are still talking, so the text is ready the moment you let go of the key.
✍️
Cleans up, never makes things up
The Voice Compiler smooths filler and formatting, but it cannot invent words you did not say. Your names, numbers, and code stay exactly as spoken.
📓
Learns your words, not your life
Corrections become small, local rules you can read and delete. There is no cloud transcript following you around.
🛟
Nothing gets lost
Faithful fallbacks, exact-once insertion, and a recoverable Outbox mean a bad moment never eats your sentence.
How it works
From your mouth to your cursor.
Five steps, all on your machine, in the time it takes to let go of a key.
-
Hold your key
Press and hold your hotkey, then just start talking. No window to open, no button to find.
-
It listens on-device
A warm cascade of local recognizers turns sound into text as you speak — nothing leaves the Mac.
-
It tidies the text
Light cleanup fixes filler and formatting without changing what you meant.
-
It claims the spot
Whisper Face leases the field under your cursor and double-checks it is still the right one.
-
It types
One clean paste. If the field moved, the text waits safe in your Outbox instead.
Pick your face
A little transcriptionist in your menu bar.
Choose who greets you up top. It sits quietly, then opens its mouth the moment you start talking. Hover to say hi — click to send it up to the top of the page.
Your voice never leaves the laptop.
No cloud calls. No telemetry. No transcript sitting on someone else's server. Listening, recognizing, cleaning up, and typing all happen on your Mac — even with the Wi-Fi off. When Diagnostics writes a support bundle, it stays private on your disk and contains no transcripts.
on-device recognitionzero telemetryno audio uploadlocal-only rules
Get started
Setup is part of the product.
One command provisions everything and checks its own work. Then hold your key and talk.
1
Download and open
Grab the latest build and open it. No account, no sign-up wall.
2
Run Install.command
One repeatable flow sets up the app, locked dependencies, models, the native helper, login services, and health checks.
3
Grant mic and accessibility
macOS asks once for microphone and accessibility access — that is what lets Whisper Face hear you and type for you.
Building from source today — clone the repo and run Install.command. A signed, notarized release is on the way.
# from the Whisper Face folder
you@mac ~/Whisper-Face $ ./Install.command
✓ app installed
✓ models ready (offline)
✓ native helper linked
✓ microphone + accessibility
✓ health checks passed
Hold your key and start talking → Questions
The honest FAQ.
Is it really free?
Yes. Core dictation, privacy, accessibility, the installers, and local use are free and stay that way. No account, no word cap, no paywalled accuracy.
Does my voice go to the cloud?
No. Recognition runs on your Mac. You can turn off Wi-Fi and keep dictating. Nothing about what you say is uploaded.
Which Macs does it run on?
Modern Apple Silicon Macs. The installer checks your machine and models during setup and tells you if anything is missing.
What about Windows or iPhone?
Windows shares the same core pipeline and keeps one-click installer parity. A native iPhone product is a future workstream, not a setup promise today.
How accurate is it, honestly?
It runs multiple local recognizers and protects names, numbers, code, and commands rather than smoothing them into something plausible. The repo includes a dated, neutral competitor evaluation instead of cherry-picked claims.
Is it open source?
Yes — source-available, with a separate commercial license option. The full license and third-party notices live in the repository.