Relationship of perceived emotional response to the soundscape and urban green space based on a deep learning approach

Abstract

While urban greenspace is widely recognized as important for human health and well-being, research on this topic of urgent importance is regularly not scaled for landscape planning and design applications. Here we propose a spatially explicit, deep-learning-based method to assess auditory stimuli within an important area of healthcare supply and we predict and map emotional responses to its soundscapes. Decomposing soundscape emotion by its dimensions of pleasantness and eventfulness, we find that both pleasantness and eventfulness are significant correlated with greenspace. Pleasantness is positively associated with greenspace whereas eventfulness is negatively associated with greenspace. The direction of emotional response to urban greenspace comports with current understandings of restorative landscapes. Our findings indicates that restorative soundscapes for hospitals may be insufficient due to potential impacts of urban contexts. Our spatially explicit framework helps to inform understandings of landscape restorativeness at the landscape planning scale and can be replicated to help ensure landscape design reaches its restorative potential, particularly in critical urban applications where populations and public health needs are increasingly concentrated.