SenUrbanNoise: An Urban Sound Event Dataset for Acoustic Monitoring in Senegal

SenUrbanNoise is an urban sound event dataset developed for acoustic monitoring and machine learning research in Senegal. The original corpus consists of 800 audio recordings distributed across 20 sound event classes, with 40 recordings per class. Data augmentation techniques, including pitch shifting, Gaussian noise addition, amplitude modification, and reverberation, were applied to expand the dataset to 5,600 audio recordings, with 280 recordings per class. The dataset covers a diverse range of urban, human, animal, transportation, and environmental sound events. It is intended for research in environmental sound classification, urban noise monitoring, machine learning, deep learning, and AI-based acoustic surveillance.