The data and the R code used in the paper named "How multisensory integration shapes perceived comfort in urban soundscapes"

Our work addresses a fundamental yet understudied question: How do auditory and visual inputs interact with personal factors to influence comfort and well-being in noisy urban environments? Using panoramic audiovisual recordings from 31 urban streets in Taipei and perception data from 252 participants, we combined deep learning–derived environmental features (SegFormer, YOLOv8m, YAMNet) with a two-stage analytical framework integrating Generalized Additive Mixed Models (GAMM) and Structural Equation Modeling (SEM). This approach enabled us to capture nonlinear and threshold effects in sensory perception while revealing structural pathways between environmental, perceptual, and psychological factors.