Abstract

The urban noise environment comprises many sources, some of which are regulated by local legislation setting maximum permitted noise levels, which are vital in implementing the noise action plans. A multidisciplinary project funded by the Chilean R+D Agency has resulted in a machine learning-based system called FuSA that automatically recognizes sound sources in audio files recorded in the urban environment to assist in their analysis. FuSA (Integrated System for the Analysis of Environmental Sound Sources) incorporates a deep neural model transferred to a dataset of urban sound events compiled from public sources and recordings. The target dataset follows a customized taxonomy of urban sounds. The system also uses a public API so potential users can post audio files to determine the overall presence of noise sources contributing to environmental noise pollution. This work provides examples of how stakeholders can use FuSA to address urban noise problems and contribute to city noise abatement policies.