Audio-based early warning system of vehicle approaching event for improving pedestrian's safety

摘要

This paper presents an audio-based early warning system of vehicle approaching event for improving pedestrian's safety. Sound signals were collected by an external directional microphone connected to a smart phone. Multiple feature techniques, such as root mean square, zero crossings, spectral centroid, and spectral rolloff, were applied to short-time frames of audio samples. Multiple machine learning classifiers, such as K nearest neighbor, multi-layer perceptron, decision tree, and random forest, were applied to classify the audio frames to effectively detect vehicle approaching sound. The experimental results showed that the accuracy of the system is as high as 97% for two class (No Vehicle and Vehicle) classification.