Analysis and automatic detection of anomalous noise events in real recordings of road traffic noise for the LIFE DYNAMAP project

The LIFE DYNAMAP project envisions the development of an anomalous noise event detection algorithm that aims at excluding non-road traffic noise sources from the sound pressure levels represented on the dynamic noise maps. With the aim of adapting the algorithm to the real acoustic environment of the project pilot areas, a recording campaign was conducted. As a result, multiple samples of background noise and road traffic noise together with anomalous noise events under different weather conditions were obtained. In this work, we present an in-depth analysis of the recorded database, including a study of the distribution of all the types of collected anomalous noise events, which are also analyzed in terms of their duration and signal-to-noise ratio. Taking into account the conclusions drawn from this real world data analysis, we implement a supervised anomalous noise event detection algorithm, and evaluate its performance in one of the project pilot areas, testing not only its ability to detect anomalous noise events, but also its sensitivity in terms of the acoustic salience of the events with respect the surrounding traffic noise.