Automatic classification of urban traffic noise onboard an acoustic monitoring system

Recent developments in acoustic monitoring systems make it possible to measure complex noise situations, like urban traffic noise, continuously. Monitoring provides more insight in the noise situation, from which more specific and (cost) effective measures can be taken. Monitoring also allows direct evaluation of the effect of mitigation, or to act directly upon sudden unwanted noise situations. However, continuously monitoring with high sample rates results in large quantities of data, which need to be processed to obtain the relevant information. Intelligent onboard processing is an efficient way of dealing with this issue. Additionally, onboard processing results in less data communication which is often a of practical relevance. In this paper a classification module is presented that is configured for usage during urban traffic noise monitoring. The module is designed to help manage the amount of data by marking areas of interest by assigning classes to the recorded audio. The classification module thus adds additional value to the collected data. A monitoring system equipped with the module could, for example, automatically publish reports describing multiple aspects of the sound