Automatic classification of traffic noise

Abstract

When modeling a city or a secondary road to calculate a noise map, the information about the number of heavy/light vehicles and the average speed it is not always available. In this paper, a first approach to get an automatic classification of vehicles is presented. The system is based on the classification of the audio signal that a noise source produces. Some basic classifiers have been tested (k-nearest neighbours, FLD (Fischer linear discriminator) and principal components. As first approach, the aim of the job was to determine if the different classes (trucks, cars, and motorbikes) could be separable using different time and frequency characteristics: zero crossing ratios, spectral centroids, spectral rolloff, subband energies and mel frequency cepstral coefficients. The results shows that for some of the characteristics tested, the signals are separable, so a continuous traffic noise signal could be processed to get the information of the number of heavy trucks, cars, and motorbikes that passed by during the recording time. Information of a stereo recording could be used to get information of the direction of the vehicle. At this moment, combining three characteristics and FLD, errors bellow 9% can be reported.