Troubleshooting
Start with the menu
Click the TongueType icon in the menu bar. When something is wrong, the top of the menu usually says so. A red row reading “Microphone access needed. Click to enable.” or “Input Monitoring needed. Click to enable.” means a permission is missing; clicking it shows the macOS prompt or opens the right pane in System Settings. A Finish Setup… entry means setup never completed and re-opens the wizard. The status line also reads “Loading model…” or “Speech model failed to load” when the model is the problem.
If the menu looks normal, find your symptom below.
The hotkey does nothing
Input Monitoring is missing. This is the usual cause. Without that permission, macOS never tells TongueType you pressed the key, and there is no error to show. Click the red row in the menu, or go to System Settings → Privacy & Security → Input Monitoring and toggle TongueType on.
Expect TongueType to quit and relaunch on its own when you flip that switch. That is macOS behavior, not a crash. The app is back within a second or two and the setup wizard may reappear so you can confirm everything works.
Wrong key, or a key with a day job. Check which hotkey is selected under Settings… → Hotkey. If you chose the Globe key, set “Press Globe key to” → Do Nothing in System Settings → Keyboard, or macOS opens the emoji picker instead. If you chose a function key, enable “Use F1, F2, etc. as standard function keys” in the same place. Keyboard remappers like Karabiner or BetterTouchTool can also swallow a key before TongueType sees it; switch to a different hotkey to rule that out.
You tapped instead of holding. TongueType ignores presses shorter than the grace period (100 ms by default) so brushes against the key do nothing. Hold the key while you talk, or turn on Double-tap to latch to record hands-free.
TongueType isn’t running. There is no Dock icon, so this is easy to miss. Look for the icon in the menu bar near the clock. If it isn’t there, open TongueType from your Applications folder, and consider turning on Automatically start at login under Settings → General.
“Microphone access denied”
The overlay shows this when you press the hotkey without Microphone permission, and System Settings opens to the Microphone pane at the same time. Toggle TongueType on there and press the hotkey again.
If TongueType isn’t listed in that pane at all, macOS has recorded a decision without showing it. Quit TongueType, open Terminal, and run:
tccutil reset Microphone net.abeautifulsite.TongueTypeRelaunch TongueType and press the hotkey. The permission prompt appears as if for the first time.
“Microphone unavailable” or “Microphone disconnected”
The microphone you picked went away mid-session, or macOS couldn’t open it. TongueType falls back to the system default automatically and reconnects to your choice when it comes back. If you’d rather not wait, pick another device under Microphone in the menu. The System Default entry shows the device name in parentheses so you can see what macOS is actually using.
A greyed-out “Microphone unavailable on this Mac.” row at the top of the menu means macOS couldn’t open any input device even though permission is granted. Check System Settings → Sound → Input for a working device. Bluetooth headsets sometimes connect for output only; disconnect and reconnect them, or try the built-in microphone. If the row stays after you’ve fixed the hardware, quit and relaunch TongueType.
“Loading model…” for a long time
The Whisper model takes a minute or two to load onto the Neural Engine after each launch. TongueType starts that load in the background the moment the app opens, and plays a soft chime when it’s ready (if sounds are on). The first launch after a macOS update can take a little longer.
If the status line reads “Speech model failed to load,” quit and relaunch. Make sure the Mac has a couple of gigabytes of free disk space, since macOS compiles the model on first load. If it fails again, choose Finish Setup… from the menu and click Retry. Still stuck? See the last section on this page.
Text goes to the clipboard instead of my cursor
If the overlay says “Copied · press ⌘V to paste,” the optional Accessibility permission isn’t granted. TongueType has fallen back to putting the text on the clipboard, which is the designed behavior, not a bug. To insert directly, toggle TongueType on under System Settings → Privacy & Security → Accessibility. If the toggle is already on but you still get the clipboard, flip it off and on again to refresh macOS’s record.
If the overlay says “Copied to clipboard” with no ⌘V hint, your Output mode is set to Copy text to clipboard under Settings → General. Switch it to Insert text at cursor if that’s not what you wanted.
Also check that a text field is focused. If nothing on screen can accept typing, there’s nowhere for the text to go.
Capitalization or spacing is off
Smart insertion reads the text around your cursor to decide whether to start lowercase or add a space. Some apps don’t expose their text to macOS, especially terminals and a few code editors, so there TongueType inserts the transcript exactly as Whisper wrote it. If you’d rather always get the raw output, turn off Smart insertion under Settings → General.
Two things that look like bugs but aren’t: Caps Lock on while you dictate produces ALL CAPS on purpose, and password fields are never read or adjusted, so no smart adjustments happen there.
It transcribed the wrong language, or gibberish
Check Language in the menu. If it’s set to a language you weren’t speaking, Whisper does its best to fit your words into it, with strange results. Pick the right one. Auto-detect works too, but it guesses from the first few words and can miss on short phrases, so choosing a language explicitly is more reliable. The multilingual dictation page has more on switching.
Gibberish from a near-silent recording is a Whisper quirk: given almost nothing to hear, it sometimes invents text. Move closer to the microphone or pick a better one under Microphone in the menu.
“[music]” and “(laughter)” show up in my text
Whisper describes non-speech sounds when it hears them. Turn on Filter non-speech annotations under Settings → Postprocessing to strip them before the text reaches your cursor. It’s off by default so nothing you actually said gets eaten.
I said “new line” and got the words “new line”
Make sure Replace spoken phrases is on under Settings → Postprocessing, and that the rule itself is checked in the list. The defaults are English phrases, so if you dictate in another language you’ll want to add your own, for example “nueva línea” → \n. See Postprocessing for how matching works.
My recording vanished
Ending a recording with “scratch that,” “cancel that,” or “discard that” throws the whole thing away on purpose. So does pressing Esc while recording. If a cancel phrase fires when you didn’t mean it, edit the list or turn off Cancel transcription with a keyword under Settings → Postprocessing.
With Double-tap to latch on, remember that tapping the hotkey again stops and transcribes, while Esc discards.
“Monthly limit reached”
The free tier includes 30 minutes of live dictation per calendar month. The meter resets on the 1st, and the menu shows how much you’ve used. Files are trimmed to their first 10 seconds on the free tier regardless of the meter. TongueType Pro lifts both limits with a one-time purchase; see Free vs Pro.
The setup wizard keeps coming back
The wizard re-opens at launch while either Microphone or Input Monitoring is missing, because live dictation can’t work without them. Grant both and it stops. Closing it with the X dismisses it for the current session. The optional Accessibility permission never triggers it.
I can’t find the menu bar icon
It’s probably hidden behind the notch, pushed off-screen by a full menu bar, or tucked into a menu bar manager like Bartender or Ice. Make the frontmost window narrower or quit a few other menu bar apps. Pressing the hotkey confirms TongueType is running even when you can’t see it. Getting Started has the full rundown.
Settings didn’t sync to my other Mac
Sync settings via iCloud is a per-Mac toggle under Settings → General, so turn it on on both machines, both signed into the same Apple ID with iCloud Drive enabled. The first Mac to turn it on asks whether to use the settings already in iCloud or replace them. A few things stay local on purpose: your microphone choice, the transcribe-file shortcut, lifetime stats, and permission grants. Details are in Settings.
My license key won’t unlock Pro
Paste the key exactly as it appears in your email, with no extra spaces. One license covers 5 Macs; if they’re all in use, open Unlock Pro… on a Mac you no longer use and click Deactivate this device. Lost the key? Click Lost your key? in the same window, or sign in to your Polar customer portal with the email you used at checkout. See Free vs Pro for the details.
Still stuck?
Try Check for Updates… first, since fixes ship often. Then head to Support and tell us your macOS version, which Mac you have, what the top rows of the menu say, and what you did right before it went wrong. Screenshots help.
Where to next
- Permissions: what each of the three macOS permissions does, and how to revoke or reset them.
- Getting started: the setup wizard, step by step.
- Support: when the page above didn’t help.