A system and method for detecting speech from occupants in a building management system is disclosed. Building management systems include fire alarm systems, building automation systems and security systems, in examples. Installed devices deployed within the building include audio transducers that detect speech from the occupants, and a management system panel processes the information from the installed devices and processes the detected speech from the occupants. In a fire alarm system, in one example, the fire alarm panel processes the detected speech from fire sensor devices and alarm notification devices as the installed devices. The fire alarm panel and/or its installed devices can identify commands from the detected speech for controlling and testing the fire alarm management system. In embodiments, Artificial Intelligence (AI) subsystems can be further added to the building management systems for control and information services.