Troubleshooting MojoFlow
Fix dictation that types nothing, the wrong words, or in the wrong place, and problems with updates.
When a dictation fails, MojoFlow says why on the Overview page and in the tray icon's tooltip, even if the window was closed at the time. Start there.
- Is the microphone muted? MojoFlow tells you when the microphone is muted. Many laptops have a key that mutes the microphone; your system's sound settings can unmute it too.
- Is it the right microphone? The Overview page shows the microphone the next dictation will use. Change it under Dictation → Microphone, then hold your key and watch the level bars move.
- Bluetooth headset? It can take a second to wake up, so a short dictation can come out empty. Wait a moment after pressing the key before speaking, or use a built-in or wired microphone.
- Is dictation paused? A grey tray icon means paused: choose Resume dictation in the tray menu.
The speech model heard silence or background noise and guessed. MojoFlow filters most of these out; if they still appear, check the microphone is the one you are speaking into and that its level bars move when you talk.
Add it to the dictionary. For a name that only needs its spelling kept, fill in Should be alone. Setting Spoken language to your language rather than Auto-detect also helps.
- Check the key under Dictation → Push-to-talk → Key, and that MojoFlow is running (its icon is in the tray).
- Overview says "One download to go" until a speech model is downloaded.
- On Linux, a single key such as Right Ctrl needs the keyboard rule offered by the installer. Run the install command again and accept it, or choose a combination such as Ctrl + Space.
- On Windows, MojoFlow cannot type into an app running as administrator (Windows does not allow it from an ordinary app).
MojoFlow types where your cursor is when you let go, so keep the app you are dictating into in front. If an app drops characters while MojoFlow types, set Dictation → How text is inserted to Paste (Windows). On Linux, check Dictation → Permission to type: if it is not allowed, choose Allow typing and accept your desktop's Remote Desktop window.
Clean-up is off or its model is not downloaded. Open Clean-up: set Mode to Light and download the model under Model status. If a toast said "Clean-up was skipped this time", that one dictation was typed as heard and the next will be cleaned up.
MojoFlow cannot see which app is in front, and the Clean-up page says why. On GNOME, sign out and back in once after installing so the helper extension starts, and make sure extensions are switched on in the Extensions app.
The Speed card on the Overview page shows the typical wait. The first dictation after MojoFlow starts loads the models and takes longer. Check Models → Graphics chip: on a laptop with two, the separate graphics card is usually faster. The battery saver model is slower by design. A smaller speech or clean-up model is faster at some cost in accuracy.
MojoFlow never runs an installer it cannot verify, and About says why it refused. Choose Try again, or Download manually to get the installer from the Mojo Up portal. See Updates.
Sending a problem to Mojo Up
MojoFlow records anything that goes wrong in errors.log, in %LOCALAPPDATA%\MojoFlow on Windows or ~/.local/share/MojoFlow on Linux. Attach it, with your version (bottom of the sidebar), when you contact support.