Keywords: soundscape, semantic annotation, sound phenomenology, audio synthesis, semiotics of the audible
This paper presents a formalised model for the dynamic generation of soundscapes and its implementation in an application that allows for an interactive, real-time exploration of the resulting soundscapes. After discussing the notion of soundscape and its respective literature, a formalizing definition is proposed that takes into account three main types of sound objects: atmospheres, events, sound subjects. The resulting soundscape model is integrated into a methodology that aims at generating synthetic soundscapes from existing ones by means of a human-based annotation process. The methodology features four steps: classification, analysis and recording; production; generation; evaluation. This synthesis-by-analysis model is applied to the simulation of the soundscape of the Porta Palazzo market in Turin.