Ambulance Siren Detection using an MFCC based Support Vector Machine

This paper proposes a mechanism to accurately identify ambulance sirens using an audio signal processing technique and an ML algorithm to provide consistent predictions. The improvement in the accuracy of the machine learning model can be attributed to the improved training dataset. Further, since the training dataset can cause a higher computational overhead, this paper proposes a method to achieve a reduction in the dimensionality of the training dataset. The dataset obtained from the Kaggle platform has been compared with the dataset that is created specifically for this study. This comparative study is used to demonstrate the type of dataset that produces the most efficient prediction of ambulance sirens.