An intelligent sound alam recognition system for smart cars and smart homes

Our work describes a unique alarm recognition scheme for smart cars and smart homes that recognizes several alarm sound categories such as a police alarm or ambulance alarm while driving, or smoke or gas alarms while at home. This system is essential in emergency situations on the road or at home. Our algorithm uses Short-Time Fourier Transforms and GPS data to analyze a small section of the signal and compare it to a pre-defined list of analyzed alerts that we maintain. Compared to human alarm recognition while driving survey results, our scheme achieves much better results against the same background noise, more than doubling the human recognition distance.