Calibrate for 10 seconds while the room is quiet, especially after moving the microphone. Calibration sets the minimum volume for the selected activity mode; it does not learn talking as background noise.

Manage modes to lower Min Vol if quiet talking is missed. Chatter % is the allowed share of speech over the last 10 seconds; lower it to respond to less talking. The speech model runs locally after its files download. If it cannot load, the status above explains the frequency-based fallback, which can mistake other sounds for voices.

Use Pause while the teacher speaks. Pause and Resume remain available on the main screen when the panel is hidden. Resuming discards old audio samples so paused speech is not counted later.

Select a class before changing its bar. Each class has its own bucket size, set in Manage Classes (new classes start at 500). + and - change the bar by 1% of the class's bucket size; Off-Task adds 5%. Drag vertically with a mouse or touch, or focus the bar and use arrow keys (1%), Page Up/Down (10%), Home (empty), or End (full). A drag saves when released; canceled drags do not change the score. Adjustments sync with open pop-outs and remain after reloading.