Exploration of perceptual differences of virtually enhanced sound fields using timbre toolbox

In this presentation, we approach the analysis of perceived spatial differences from the perspective of timbre using the Timbre Toolbox [Peeters et al., 2011, J. Acoust. Soc. Am., 130, 2902–2916]. The Timbre Toolbox, using MATLAB, calculates 88 descriptors that have been reported in various timbre perception literature. These include temporal (such as RMS energy), spectral (both in magnitude and power spectra), and harmonic descriptors, as well as those based on the Equivalent Rectangular Bandwidth (ERB) model, which approximates the auditory processing. Analyses were performed on 27 stimuli of nine string quartet groups performing one excerpt in three room conditions, as well as the data from perceptual experiments reported in an earlier study [Chon et al., 2015, Proc. Audio Eng. Soc.]. The three room conditions have different spatial profiles, which consist of one natural room and two virtually enhanced rooms, where many physical parameters were closely controlled (e.g., ST1 and ST2). The experimental data include both the musicians’ ratings of spatial quality for performance (e.g., tonal balance and envelopment) as well as recording engineers’ ratings of recorded spaces (e.g., clarity and room size). The best timbre descriptor for each question will be reported, and an overall merit of this approach discussed.