Isolate input#
Use a short invented sentence at ordinary speaking volume. Check CueSpec’s status overview for microphone permission and model readiness. Confirm that the device you expect is connected and available. If a headset has disconnected, do not assume the host has switched to the input you intended.
A simple recording test in another permitted audio app can help distinguish a general device issue from CueSpec’s capture path. It does not prove CueSpec itself is ready; return and test its actual workflow.
Then check transcription#
If audio is present but text is absent, examine local model readiness and the transcription state. If only whispering fails, establish normal speech first and evaluate soft capture separately.
Retain the visible error, version and input-device name for a report. Avoid sending private audio to a public support thread. A one-sentence reproduction is more useful than repeatedly recording a long request into an empty draft. Do not reinstall or reset account data before locating the failure stage.