Dictation: speak, and it types at your cursor
On this page
Double-tap fn, say what you want to type, and tap fn again. Short Leash turns what you said into text and pastes it where your cursor is: your editor, a terminal, a chat, any app. The speech model runs on your computer. No audio is saved, and nothing is sent anywhere.
Dictation is part of Short Leash Pro.
Set it up
Settings › Dictation › Set up dictation walks through it. Its first step shows the speech model (Parakeet Ultra, about 740 MB, picked from the languages you speak) and Download model: nothing downloads until you choose it, and the download keeps going while you set up the rest. Setup finishes once the model is ready. It asks for:
- Microphone, so Short Leash can hear you while you dictate. Short Leash listens only after you start dictation, and a lamp shows the whole time it does.
- Accessibility (macOS), so Short Leash can paste for you. macOS may ask you to restart Short Leash after you allow it.
- A shortcut, which you then press once to check Short Leash hears it. If it hears nothing on a Mac, setup asks for Input Monitoring too.
On a Mac whose Press 🌐 key to setting shows emoji or changes input source, tapping fn does that as well, so setup and Settings › Dictation ask you to set it to Do Nothing in Keyboard settings. Not set at all is macOS's default, which shows emoji. Short Leash reads that setting; it never changes it.
Shortcuts
| Start | Stop | |
|---|---|---|
| Double-tap fn (Mac, the default) | Double-tap fn | Tap fn |
| Hold fn | Hold fn while you speak | Let go |
| Control+Option (keyboards without an Apple fn key) | Double-tap, or hold | Tap, or let go |
| Windows | Hold Ctrl+Win while you speak | Let go |
| Windows, hands-free | Ctrl+Win+Space | Ctrl+Win+Space |
esc cancels and throws away what you said. A shortcut with another key added, like fn+Delete, fn+arrows or Ctrl+Win+Enter for Narrator, never starts dictation: those keys go to the app or the system as usual. Hands-free dictation stops by itself after a minute of silence.
The menu bar (the tray on Windows) has a Dictate pill, and Stop dictation while Short Leash listens.
Your own shortcut
When something else already uses fn or Ctrl+Win, record your own in Settings › Dictation: choose Record shortcut and press it. It can be:
- a key with Control, Option (Alt) or Command (Win), like ⌃⇧D or Ctrl+Alt+D;
- a function key alone, like F13;
- one modifier on its own, on its side, like right ⌥ or right Alt;
- two or more modifiers together, like ⌃⇧.
A letter, number, Space or arrow alone or with only Shift would get in the way of typing, so Short Leash asks you to add a modifier. Escape and Caps Lock can't be used. Hold to talk works with your own shortcut on both systems: double-tap it to start and tap it to stop, or hold it while you speak.
Before you save, Short Leash names the shortcuts it would collide with, such as Spotlight's ⌘Space, macOS Dictation's Press Right Command Key Twice, Windows' Win shortcuts, Narrator's Ctrl+Win+Enter and the On-Screen Keyboard's Ctrl+Win+O, and copy and paste in every app. You can still use it; both would happen. Nothing is saved until you choose Use this shortcut, and esc cancels recording.
A shortcut with a key that isn't a modifier, like ⌃⇧D, is kept from the app in front, so pressing it doesn't type a D. On a Mac that needs Accessibility as well as Input Monitoring. A modifier alone on Windows, like right Alt or Win, stops tapping it from showing apps' menu keys or opening Start, and Short Leash says so when you record it.
Where the text goes
The text is pasted where your cursor was when you stopped. A small pill near the bottom of the screen says what happened: Pasted into Slack, or why it couldn't paste.
Short Leash puts the text on the clipboard instead, and tells you to press ⌘V (Ctrl+V), when:
- you moved to another app while the text was being made;
- another app has secure typing on, as password managers do while you type a password;
- the field is a password field, or the window runs as administrator (Windows).
To undo a dictation, use ⌘Z (Ctrl+Z) in the app you dictated into. The menu bar keeps your last 10 dictations, in memory only: the button beside Dictate shows the last one, with Paste last dictation and Copy last dictation. They're gone when Short Leash quits.
Microphone
Short Leash uses your computer's own microphone unless you choose another in Settings › Dictation. Bluetooth headphones like AirPods switch to a lower sound quality while their microphone is on, so Short Leash leaves theirs alone and says so once, with Use AirPods instead.
Languages
Languages I speak in Settings › Dictation picks the speech model: Parakeet Ultra covers English and 24 other European languages. Other languages use other models, which Settings › Models lists with their sizes.
If it doesn't work
- "Can't hear the microphone": check that Short Leash is allowed in Privacy & Security › Microphone (Settings › Privacy & security › Microphone on Windows), and that the microphone in Settings › Dictation is the one you're speaking into.
- Nothing happens when you press the shortcut: run Test the shortcut in Settings › Dictation. On a Mac, allow Short Leash in Privacy & Security › Input Monitoring and restart Short Leash; your own shortcut with a key like ⌃⇧D needs Accessibility too.
- fn or Ctrl+Win does something else as well: record your own shortcut in Settings › Dictation.
- It's always copied instead of pasted: allow Short Leash in Privacy & Security › Accessibility, then restart Short Leash.
To remove the permissions macOS gave Short Leash, run tccutil reset All dev.shortleash.app in Terminal.