Expression communication is the added value of a musical performance. It is part of the reason why music is interesting to listen to and sounds alive. Previous work on the analysis of acoustical features yielded relevant features for the recognition of different expressive intentions, inspired both by emotional and sensorial adjectives. In this article, machine learning techniques are employed to understand how expressive performances represented by the selected features are clustered on a low-dimensional space, and to define a measure of acoustical similarity. Being that expressive intentions are similar according to the features used for the recognition, and since recognition implies subjective evaluation, we hypothesized that performances are similar also from a perceptual point of view. We then compared and integrated the clustering of acoustical features with the results of two listening experiments. A first experiment aims at verifying whether subjects can distinguish different categories of expressive intentions, and a second experiment aims at understanding which expressions are perceptually clustered together in order to derive common evaluation criteria adopted by listeners, and to obtain the perceptual organization of affective and sensorial expressive intentions. An interpretation of the resulting spatial representation based on action is proposed and discussed.

During a music performance, the musician adds expressiveness to the musical message by changing timing, dynamics, and timbre of the musical events to communicate an expressive intention. Traditionally, the analysis of music expression is based on measurements of the deviations of the acoustic parameters with respect to the written score. In this paper, we employ machine learning techniques to understand the expressive communication and to derive audio features at an intermediate level, between music intended as a structured language and notes intended as sound at a more physical level. We start by extracting audio features from expressive performances that were recorded by asking the musicians to perform in order to convey different expressive intentions. We use a sequential forward selection procedure to rank and select a set of features for a general description of the expressions, and a second one specific for each instrument. We show that higher recognition ratings are achieved by using a set of four features which can be specifically related to qualitative descriptions of the sound by physical metaphors. These audio features can be used to retrieve expressive content on audio data, and to design the next generation of search engines for music information retrieval.

This paper reports on a study on the perception and rendering of distance in multimodal virtual environments. A model for binaural sound synthesis is discussed, and its integration in a real-time system with motion tracking and visual rendering is presented. Results from a validation experiment show that the model effectively simulates relevant auditory cues for distance perception in dynamic conditions. The model is then used in a subsequent experiment on the perception of egocentric distance. The design and preliminary result from this experiment are discussed.