Soundscape Fingerprinting - Methods and Parameters for Acoustic Assessment
Soundscape, as defined in ISO 12913-1, is understood as a perceptual construct whose evaluation is recommended by a triangulation between the components person, context and acoustic environment. With the increase of information about each of these components, the description of a specific soundscape gains validity. This paper is motivated by this approach and focuses on the identification of methods, parameters and indicators for the acoustic description of soundscapes. The proposed methodology goes along and extends the recommendations from ISO 12913-1/-2/-3 and is based on an analysis of a large database of spatial (Ambisonics) recordings that covers a wide range of anthropogenic and natural soundscapes. A signal analysis is conducted in terms of parameters of the acoustic categories of timbre, level, sound sources, spatiousness and temporal behavior. A subsequent dimensionality reduction using multivariate statistics is used to extract those parameter combinations that best characterize a soundscape from an acoustic aspect and that therefore contribute to the development of a discriminative acoustic soundscape fingerprint.This contribution presents the methodology and implementation as well as intermediate results. Possible applications of such fingerprints can be manifold: e.g. (un-)pleasantness modeling, acoustic scene classification or the validation of reproduced immersive soundscapes using loudspeakers or headphones.