Getting started
Sayzen needs a Mac with Apple silicon and macOS 14 or later, or a PC with 64-bit Windows 10 or 11. Setting it up takes a few minutes.
Install
Download Sayzen. On a Mac, open the disk image and drag Sayzen to Applications. Sayzen isn’t notarized by Apple yet, so the first time you open it macOS shows “Sayzen” Not Opened. Choose Done, then open System Settings, go to Privacy & Security, scroll down and choose Open Anyway.
On Windows, open Sayzen-Setup.exe. When Windows shows “Windows protected your PC”, choose More info, then Run anyway.
Allow permissions
On a Mac, Sayzen asks for these in System Settings › Privacy & Security:
- Microphone, to hear you while you dictate or ask.
- Accessibility and Input Monitoring, so fn and side-specific keys work as shortcuts in every app, and finished text can be typed into the field you were in.
- Screen Recording, optional. With it, Commands can see your screen. Annotate asks for it the first time you use it.
- System Audio Recording, asked the first time you record a meeting, so Sayzen can hear the call’s sound.
If a shortcut does nothing, check that Sayzen is turned on in Accessibility and Input Monitoring. On Windows, allow microphone access in Windows Settings › Privacy & security › Microphone.
Download a speech model
Speech is turned into text on your computer. Sayzen starts with Parakeet, NVIDIA’s model for English: fast and accurate, and downloaded once when you set Sayzen up.
For other languages, choose a Multilingual model in Settings › AI. These are Whisper models: larger ones are more accurate and need more disk space and memory. Then set Settings › Dictation › Spoken language to Any language, and Sayzen finds the one you speak. Parakeet hears English only.
Dictation cleanup
Sayzen tidies what you said: punctuation, filler words, lists. On a Mac with Apple Intelligence (macOS 26 or later), this runs on the Mac with Apple’s own model. It’s free and needs no key.
You can have your AI provider clean up instead, in Settings › AI › Dictation cleanup. Where Apple Intelligence can’t run, and on Windows, cleanup uses your provider when a key is saved; without one, your words stay as transcribed.
Add an API key
Dictation works without a key. Commands, the Assistant and asking about a note use an AI provider, with your own key:
- In Settings › AI, choose Anthropic, OpenAI or OpenRouter.
- Paste your key from that provider and save it. On a Mac it is stored encrypted with a key kept in the macOS Keychain; on Windows it is sealed on the PC.
- Pick a model. Each provider keeps its own model choice.
Shortcuts
| Action | Mac | Windows |
|---|---|---|
| Hold to dictate | fn | Right Ctrl |
| Hands-free dictation | fn and Space | Right Ctrl and Space |
| Hold to ask (Commands) | Right Command | Right Alt |
| Talk with the Assistant | Right Command, tapped twice | Right Alt, tapped twice |
| Paste last dictation | Control-Option-V | Alt+Shift+V |
| Annotate the screen | Control-Option-A | Ctrl+Shift+F8 |
| Stop recording | Enter, or say “enter” | Enter, or say “enter” |
| Cancel | Escape | Escape |
When a Command’s answer is showing, Enter pastes it into the app you’re in. You can turn that off in Settings. Paste last dictation types your last dictation again, or copies it when no text box is focused. On a Mac, Command and comma opens Settings.
Every shortcut can be changed in Settings › Dictation › Shortcuts, including left- or right-side modifier keys.
Your first dictation
Click into any text field, hold fn (Right Ctrl on Windows) and say a sentence. Let go, and the text appears in that field. With no text field selected, it goes to your clipboard. Hold Right Command (Right Alt on Windows) and ask a question to try Commands.