Facial expression portrait: A non-verbal and interactive method for soundscape assessment in urban open spaces

The soundscape in urban open spaces is essential for enhancing the livability of cities. Current soundscape assessment methods are increasingly incorporating non-verbal forms of expression; however, limitations in interactivity and flexibility persist. This study proposes the Facial Expression Portrait (FExP) as a non-verbal and interactive method for soundscape assessment. Six typical sounds from urban open spaces were chosen as stimuli in the experiment. Participants expressed their sound perception by modifying the facial expressions of virtual avatars, while their real facial expressions and verbal assessment results were also recorded. This study examines the effects of different auditory stimuli on FExP and evaluates its validity by comparison with traditional assessment methods. The results show that different stimuli have significant effects on the valence (χ² = 135.00, p<0.01), arousal (χ² = 22.61, p<0.01), and the intensities of 16 facial action units in FExPs. Compared to real facial expressions, FExPs exhibit more pronounced emotional responses. Additionally, it shows strong predictive power for verbal indicators, such as overall assessment and pleasant (R_m²>0.4). FExP provides notable advantages in terms of flexibility in data collection and the quality of visualizations. It complements Facial Expression Recognition across multiple dimensions and shows potential for mapping onto verbal indicators. This method provides new insights into soundscape assessment in urban open spaces, with potential to expand traditional assessment frameworks and support cross-disciplinary applications.