In this paper we address automatic vehicle and engine identification based on audio information. Such data depend on many factors, including vehicle type, tires, speed and its change, as well as road type. In our previous research we designed a feature set for selected vehicle classes, discriminating pairs of classes. Later, we decided to expand the feature vector and find the best feature set (mainly based on spectral descriptors), possibly representative for each investigated vehicle category, which can be applied to a bigger data set, with more classes. The experiments were performed first on on-road recordings, and then continued with test bench (dyno) recordings. The paper also shows problems related to vehicles classification, which is detailed in official documents by national authority for issues related to the national road system, but simplified for automatic identification purposes. Experiments on audio-based vehicle type and engine type identification are presented and conclusions are shown.

In this paper, we focus on detection of speed changes from audio data, representing recordings of cars passing a microphone placed near the road. The goal of this work is to observe the behavior of drivers near control points, in order to check whether their driving is safe both when approaching the speed camera and after passing it. The audio data were recorded in controlled conditions, and they are publicly available for downloading. They represent one of three classes: car accelerating, decelerating, or maintaining constant speed. We used SVM, random forests, and artificial neural networks as classifiers, as well as the time series based approach. We also tested several approaches to audio data representation, namely: average values of basic audio features within the analyzed time segment, parametric description of the time evolution of these features, and parametric description of curves (lines) in the spectrogram. Additionally, the combinations of these representations were used in classification experiments. As a final step, we constructed an ensemble classifier, consisting of the best models. The proposed solution achieved an accuracy of almost 95%, without mistaking acceleration with deceleration, and very rare mistakes between stable speed and speed changes. The outcomes of this work can become a basis for campaigns aiming at improving traffic safety.