An occupancy tracking device configured to receive sound samples, to identify voices within the sound samples, and to determine a first occupancy level based on the identified voices. The device is further configured to identify user devices connected to an access point and to determine a second occupancy level based on the user devices that are connected to the access point. The device is further configured to measure a signal strength of a network connection with the access point and to determine a third occupancy level based on the signal strength of the network connection with the access point. The device is further configured to determine a predicted occupancy level based on the first occupancy level, the second occupancy level, and the third occupancy level and to control a Heating, Ventilation, and Air Conditioning (HVAC) system based on the predicted occupancy level.