Ambulance Siren Detection using ANN

Hearing and seeing accidents and deaths happening all over the world due to poor road networking and overriding of traffic rules is just tragic. And these accidents involve an ambulance rushing to nearby hospital in the hope to save the individual’s life, and what they undergo on the way is that the congestion of traffic signals. Now the major setback found in this unhealthy situation is that the traffic police couldn’t identify the direction of ambulance with only its siren, and they could only intervene and change the light signals only when they see the ambulance. This unawareness of ambulance’s direction gave an opportunity to propose a solution to solve the problem by assisting in existing traffic light concept with the modification in its generic system by making use of the ambulance’s siren feature. Machine Learning is now permeable in every product or solution that exists, and it really does its job on helping us and it has also given us a way to solve the existing traffic problem during emergency situations making the traffic light system automatically analyse the arrival of Emergency Vehicles (EV) and assist the traffic police beforehand on which direction the ambulance is arriving so that emergency measures like changing the lights to avoid congestion in the intersection can be done. By implementing Artificial Neural Networks (ANN) with Mel-Frequency Cepstral Coefficients (MFCC) feature extraction method the model achieved a 90% accuracy and accurately predicted the siren sound which was then verified by the CSV file given by the dataset- Urbansound8k. The audio data were converted into data points using MFCC and then easily mapped them with their respective classes. With ANN and MFCC, the prediction and accuracy rate are said to be better and by simple neural networks high grade feature extraction method was implemented