Detection of Ambulance Siren in Traffic
Dharma Rane, Pushkar Shirodkar, Trilochan Panigrahi, S. Mini · 2019
In this paper, a method to detect ambulance siren in a traffic using the smart phone in real time is discussed. Ambulance uses siren sound which alerts other road users which makes them to move efficiently through traffic. It may be possible that the siren sound of the ambulance is missed due to soundproofing or audio system inside the vehicles. To overcome such situations, it is suggested that the driver's mobile phone will have an app installed that will rely on the phone's micro-phone to automatically detect the siren and alert the user. The proposed method first divides the recorded audio signal into windows and then extracts features in both time and frequency domain. Then features are used to train a Bayesian regularized artificial neural network (BRANN). A new model that relies on two feature sets at a time thereby improving accuracy and decreasing possible delay is proposed and implemented. It is observed that the proposed method provides an accuracy of greater that 99 percent in simulated conditions using sound data from prerecorded audio. Further, the contribution of the ambulance sound with the other noise is also estimated.