Three times faster
You speak at around 150 words a minute; you type at 40. The email that took you five minutes now takes one — and it comes out better written.
Hold a shortcut, say what you're thinking, let go. The text stays written right where your cursor was: your email, WhatsApp, Word, the work chat, any text field on your computer.
It's not that you type slowly. It's that you think faster than your fingers can keep up, and your day drains into that gap: emails you postpone, messages left on read, ideas that go cold while you type them out. Pilot Voice removes the toll.
You speak at around 150 words a minute; you type at 40. The email that took you five minutes now takes one — and it comes out better written.
When you dictate you don't edit while you think. The idea comes out complete and in your own voice, before it cools off or you cut it short out of sheer typing fatigue.
Your hands, your neck, your back. After eight hours at the keyboard, simply saying what you need to say feels like breathing.
Nothing to open, nothing to learn. Pilot Voice lives up in the system bar and types inside whatever app you have in front of you.
On Mac, ⌥ + Space. On Windows, Ctrl + Shift + Space. Prefer another combination? You change it in two clicks.
No robot voice, no saying “comma” and “period” out loud. Talk the way you'd talk to a person — the model handles the punctuation.
The text appears written where your cursor was. Nothing to copy, nothing to paste, no window to close.
Pilot Voice isn't an editor you have to open and then copy out of. If there's a field on screen you can type into, you can dictate into it.
The inbox you used to dread empties out by talking. Long, nuanced replies, said in the time it used to take you to type “I'll get back to you”.
WhatsApp, Slack, Teams. You reply while walking across the office, as naturally as a voice note — but leaving text the other person can read and search.
The blank page stops being scary when you tell the first draft out loud. Then you edit — which is the part you're actually good at.
The long, well-explained prompts you never bothered to type, you now dictate in full. And AI answers you far better when you tell it everything.
All the recognition happens inside your own machine. No audio going up to the cloud, no third-party server listening to what you dictate, no account keeping your words. You dictate the contract figure, the patient's diagnosis or the hard conversation with an employee — and it stays with you.
The voice model downloads once and lives on your disk. Your audio is processed there and discarded: not uploaded, not shared, not used to train anything.
After the first download you can dictate on a plane, out in the field with no signal, or with the wifi down. Local doesn't go down when the cloud does.
No email to hand over, no password to invent, no subscription to cancel later. You install it, you dictate, and nobody on the other side even knows you exist.
The voice dictation you know charges by the minute because every word of yours travels to a server somebody has to pay for. Pilot Voice doesn't work that way: the model runs on your computer. Where there's no server, there's no bill — not for us, not for you.
No subscription, no minute bundles, no word limits. Dictate eight hours a day, every day: it costs exactly the same.
No round trip to a data center, no queue behind other users. You release the shortcut and the text is already there.
The local model Pilot Voice ships tuned comes from Atenea, the AI house behind Pilot. The same technology other places bill by the minute, running free on your machine.
All of it is set up and working from minute one. You don't need to touch anything — but if you like tweaking, it tweaks.
Ships with ⌥ Space on Mac and Ctrl + Shift + Space on Windows. Record any combination you like in Settings; if another app already uses it, it tells you right away.
Turbo (the recommended one, 574 MB), a maximum-quality Turbo, and a compact 264 MB one for machines with less memory. They download on their own and resume if the connection drops.
You choose which microphone it uses and watch the live input level as you speak. If the bar doesn't move, you know that mic isn't hearing you — before you dictate for half an hour into nothing.
Set the dictation language or let it detect it on its own. The app itself is in Spanish and English too.
It keeps what you dictated so you can copy it again, all on your own machine. One switch turns it off, one button empties it.
It sits quietly in the system bar and is ready when you turn the computer on. If you'd rather it didn't, you turn that off.
If you also want it to punctuate and format, you can connect your own OpenAI, Groq or Anthropic key. It's optional, off by default, and the key is stored in the macOS Keychain or the Windows Credential Manager.
Universal installer for Mac (Intel and Apple Silicon, macOS 13 or later) and an installer for Windows 10 and 11. Same dictation, same shortcut, same settings.
Direct download, no account and no card. The first time it fetches the voice model once, and a minute later you're talking.
And now for the head, after the heart: the first time, Pilot Voice downloads the voice model (about 574 MB for the recommended one) and asks for microphone permission; on Mac it also asks for Accessibility permission, which is what lets it type the text inside whichever app you're in. After that, dictation runs entirely on your machine, with no internet.
macOS: requires macOS 13 or later; the installer is universal (Intel and Apple Silicon) and is signed and notarized by Apple. Windows: Windows 10 or 11, 64-bit; the installer isn't signed yet, so Windows may show a SmartScreen warning — click “More info” → “Run anyway”.
Frequently asked questions
It is voice dictation for Mac and Windows: you hold a shortcut, speak, and the text appears written where your cursor was, in any app: email, WhatsApp, Word, your work chat.
It is free, with no account and no word limit: no subscription and no minute bundles. The voice model runs on your computer, and where there is no server there is no bill.
No. All recognition happens inside your machine: the audio is processed there and discarded; it is not uploaded, not shared and not used to train anything. After the first model download, it works without internet.
On Mac it comes with ⌥ Space and on Windows with Ctrl + Shift + Space. You can record any combination you like in Settings; if another app already uses it, it tells you right away.
A Mac with macOS 13 or later (Intel or Apple Silicon) or 64-bit Windows 10 or 11. The first time, it downloads the voice model (about 574 MB for the recommended one) and asks for microphone permission; on Mac it also asks for Accessibility permission so it can type into the app you are in.
Contact
Have a question about Pilot Voice? Write to us and we'll reply by email.