Signal Processing Approach for the Characterization and Modeling of Urban Noise: A Case Study of a City in Western Côte d'Ivoire

Abstract

Urban environmental noise is insufficiently characterized in many secondary cities in sub-Saharan Africa. This exploratory study characterizes short-duration urban sound recordings from five neighborhoods of Duékoué, Côte d’Ivoire (Kokoman, Antenne, Résidence, Belleville, and Guitrozon). Recordings were obtained with a Samsung Galaxy A13 smartphone at 09:00, 12:00, and 17:00 on Monday, Thursday, and Sunday and processed in MATLAB to derive A-weighted equivalent levels (LAeq), percentile levels (L10, L50, and L90), nuisance indices (LNP and TNI), loudness-related descriptors, spectral centroid, zero-crossing rate, and energy fractions in broad frequency bands. The processing chain emulated standard sound-level calculations; however, because the smartphone system was not traceably calibrated against a reference sound level meter and was not demonstrated to conform to IEC 61672, the reported sound-pressure levels are interpreted as approximate estimates rather than certified regulatory measurements. Across the tabulated sites, estimated LAeq ranged from 62.62 dB(A) in Antenne at 17:00 to 72.69 dB(A) in Résidence at 17:00. Temporal profiles differed among neighborhoods, while mid-frequency energy generally represented the largest share of the spectrum. The findings show that smartphone-based signal processing can provide useful comparative information on urban acoustic patterns when its metrological limitations are stated explicitly. Longer, traceably calibrated monitoring is required before formal compliance or population-exposure conclusions can be drawn.