Modeling soundscapes: A process of analysis by anticipated sonic content.

Abstract

Through the process of constructing sound maps or soundscapes, the acoustician attempts to describe a particular aural condition using noise-related metrics. Typically, raw-number acoustical data are one means of describing aural conditions; A-weighted sound levels, time-sensitive day-night levels, and octave frequency band analysis are commonly used means of attempting to describe a particular environment by using measurements of sound energy. Many other studies suggest that additional observations of the content of the sonic environment are important when developing descriptors of a given condition. This paper describes a process by which the soundscape is measured and subsequently modeled. Raw acoustical data, in combination with listener observations, are used as a basis for the construction of a probabilistic model that contains both predictions of sonic energy and information on the aural content of a given condition. This combined model is then used to describe and predict aural conditions with a greater potential range of descriptors than those based on raw-number acoustical data alone.