Systems and methods for providing spoken messages that reflect event status of one or more hazard detection systems within a smart-home environment are described herein. The messages can inform occupants in concise manner that does not overload cognitive recognition of those occupants. For example, the messages may be prioritized to limit the amount of information that is spoken and intelligently condense information in as concise a manner as possible. This may be accomplished by using one or more speaking paradigms to compile audible messages to be played back through a speaker of the hazard detection system.