Similarity and categorization of environmental sounds

Abstract

Four experiments investigated the acoustical correlates of similarity and categorization judgments of environmental sounds. In Experiment 1, similarity ratings were obtained from pairwise comparisons of recordings of 50 environmental sounds. A three-dimensional multidimensional scaling (MDS) solution showed three distinct clusterings of the sounds, which included harmonic sounds, discrete impact sounds, and continuous sounds. Furthermore, sounds from similar sources tended to be in close proximity to each other in the MDS space. The orderings of the sounds on the individual dimensions of the solution were well predicted by linear combinations of acoustic variables, such as harmonicity, amount of silence, and modulation depth. The orderings of sounds also correlated significantly with MDS solutions for similarity ratings of imagined sounds and for imagined sources of sounds, obtained in Experiments 2 and 3--as was the case for free categorization of the 50 sounds (Experiment 4)--although the categorization data were less well predicted by acoustic features than were the similarity data.

The importance of semantics in auditory representations

“…Any MDS analysis is limited to a representation of the perceptual distances between items; it does not provide any interpretation of the dimensions, and those dimensions may not have any obvious mapping. As in the MDS space of environmental sounds in Gygi et al (2007), it does not appear that there is any obvious single factor associated with each dimension in our space. What is apparent in both spaces is a clustering of semantically related items.…”

“…In examining the MDS results of Gygi et al (2007), we saw clear evidence for some semantic clustering, indicating that the MDS procedure does not necessarily provide a pure assessment of psychoacoustic space. As we discuss below, semantic influences on the MDS solution can potentially make it more difficult for us to observe semantic effects on change deafness (i.e., they make our change-deafness experiment more conservative).…”

“…First, listeners made similarity judgments for pairs of the sounds, and these similarity ratings were used as the input for a multidimensional scaling (MDS) analysis. The output of such an analysis is a perceptual space of N dimensions, with N determined by the number of dimensions that are needed to account for most of the variance in the similarity judgments (see Gygi, Kidd, & Watson, 2007, for an example using this approach). We then used this derived space to test for semantic versus acoustic effects with a new group of listeners.…”

“…In most respects, our procedures were similar to those used by Gygi et al (2007) in their MDS analysis of environmental sounds. With one exception (see below), all of the possible combinations of the 24 sounds were presented to listeners in sequential pairs, including each sound paired with itself and both orders of every pair.…”

The Role of Sound Source Perception in Gestural Sound Description

“…Later studies showed that during a free sorting task, the event causing the sound (action/interaction, type of excitation, source) was used more often than acoustic qualities of the sound, as a criterion for categorization [Marcell et al 2000]. Further to this, the event also overrided context and location [Gygi et al 2007], and overrided any shared physical properties when participants focused on the mental image generated by each stimulus [Scavone et al 2002]. When sounds were more complex like sequences of environmental sounds, the categorization was more closely related to global judgment like the absence or presence of human activity [Guastavino 2007].…”