Troubleshooting

The shortcut does nothing

See Shortcuts and permissions: check Settings → Privacy for “on but not working”, restart the app after changing a permission, and check for keyboard remapping.

The text did not appear

• Cadenza types into the field that had focus when you started. If you switch to another window while you speak, it will not type into the new one; the text is kept so you can copy it.
• Some apps do not expose a text field to macOS. Cadenza then types into the window directly, which works in many of them but is not guaranteed everywhere.
• Password fields and other secure inputs are never typed into.
• Check that Accessibility is allowed.

Recognition is wrong or missing words

• Speak at a normal pace, close to the microphone. Very quiet speech is raised automatically, but a distant microphone still hurts.
• Try another model with Compare models with my voice. The best model depends on your voice.
• Settings → Speech → Recognition settings: if the start of a sentence is cut off, move voice-detection sensitivity toward “More sensitive”.
• English words inside Chinese sentences are the weakest spot of every local model we measured.
• No punctuation? FireRedASR2 does not write any. SenseVoice does.

macOS asks for permissions or Keychain access again

Builds are signed ad hoc, so every new build looks like a different app to macOS. After you rebuild or update, allow Microphone, Accessibility and Input Monitoring again, and allow access to your saved cloud credentials if asked.

macOS says the app cannot be opened

Releases are not notarized. Open System Settings → Privacy & Security and choose Open Anyway. See Download.

Report a bug

Open an issue with the Bug report form. It asks for your macOS version, the engine, the trigger mode and which permissions are allowed.

Do not attach recordings, transcripts, API keys or unredacted logs. The log holds state, errors and text lengths, never audio or the recognized text; remove anything private from it anyway before pasting.