Soundscape perception indices (SPIs): Developing context-dependent single value scores of multidimensional soundscape perceptual quality
Andrew Mitchell, Francesco Aletta, Tin Oberman, Jian Kang
Abstract
The soundscape approach provides a basis for considering the holistic perception of sound environments in context. Whereas steady advancements have been made in methods for assessment and analysis, a gap exists for comparing soundscapes and quantifying improvements in the multidimensional perception of a soundscape. To this end, there is a need for the creation of single value indices to compare soundscape quality which incorporate context, aural diversity, and specific design goals for a given application. Just as a variety of decibel-based indices have been developed for various purposes (e.g., LAeq, LCeq, L90, Lden, etc.), the soundscape approach requires the ability to create original indices for different uses, which share a common language and understanding. Therefore, a unified framework for creating bespoke and reference single index measures of soundscape perception is proposed, allowing for different metrics to be defined in the future. This framework is based on a four-step test-target paradigm wherein a desired soundscape perception is defined as a target distribution within the soundscape circumplex, and the two-dimensional Kolmogorov-Smirnov distance is used to test an assessed soundscape against this target. Applications and implications of this framework are discussed, and a multi-objective optimisation method for empirically defining perception indices is proposed.