INTELLIGENT EMERGENCY VEHICLE SOUND CLASSIFICATION FOR PUBLIC SAFETY

Traffic congestion in urban areas can be a nightmare for emergency vehicles as they are slowed down by the traffic condition, thus putting the patients' lives at risk for whom medical attention is urgently needed. Traditional visual and acoustic identification methods, such as flashing lights and sirens often fall short due to multiple factors like driver distraction, obstruction in the line of sight either by any vehicle or building, and even the advanced soundproofing features of modern vehicles could be a reason. This research is all about designing the correct and efficient real-time system that detects and distinguishes between emergency vehicle sounds so drivers, pedestrians, and also the management systems in their vicinity have prompt recognition and reactions to those sounds. To accomplish this, the proposed solution utilizes acoustic analysis along with sophisticated, cutting-edge algorithms by applying features extraction using Mel-frequency cepstral coefficients (MFCC). A wide range of machine learning algorithms, including Random Forest (RF), Support Vector Machine (SVM), Multilayer Perceptron (MLP), Extra Trees Classifier (ETC), and AdaBoost, were trained and tested using a comprehensive dataset consisting of emergency vehicle sirens and background traffic noise. Among them, the accuracy of Random Forest classifier is the highest, which reaches 99.17%, and AdaBoost classifier has similar performance. In this way, this system uses sound-based detection to enhance emergency response, public safety, and traffic management with innovative acoustic monitoring and analysis. The implementation of the system will streamline emergency operations and improve the efficiency and safety of urban traffic.