Parametrización de la calidad acústica en entornos urbanos usando datos de densores remotos

Abstract

Urban noise has become an increasingly critical issue affecting the health and quality of life of city dwellers. In response, many cities have deployed monitoring stations to define and control environmental noise levels. This research proposes an approach for the parameterization of urban acoustic quality in Bogotá by integrating binaural sensing, machine learning algorithms, and stochastic analysis. Seven monitoring stations equipped with 3D-printed heads and calibrated microphones were deployed to collect 4,279 five-minute audio recordings. The data was processed using Inferencer and Soundmetrics applications, extracting metrics such as LEQ, IACC, and WIACC. These metrics were then used to model transitions between acoustic states using Markov chains and probability distributions. This approach allowed the characterization of spatiotemporal noise patterns in the urban soundscape, providing a solid quantitative basis for designing adaptive mitigation strategies and supporting evidence-based decision-making in urban environmental planning.