Sound Quality Evaluation of Acoustical Environments with Multiple Sources
Invited paper
FH Düsseldorf
Wednesday 3 june, 2015, 14:20 - 14:40
0.2 Berlin (90)
Abstract:
Sonic environments in everyday life are usually composed of multiple sound
sources with different acoustical characteristics. Thereby, spectral and
temporal masking effects occurring when multiple sounds interfere with each
other, as well as visual stimuli and cognitive processes like attention
focusing influence our perception and evaluation of the acoustical
environment. In order to explain annoyance reactions to combined noise
sources, various models based on the summation of loudness or sound pressure
levels have been developed recently. However, these models are limited, as
natural sounds, for example, can also have a beneficial effect in noisy
environments, although they represent additional sound sources. Against this
background, predicting human evaluation of complex acoustical environments
still poses a challenge.
To focus on fundamental evaluation mechanisms, we conducted two listening
experiments under laboratory conditions, where we investigated how the
pleasantness ratings of singular sounds affect the overall evaluation of
their respective combinations. In the course of the first study,
participants had to evaluate technical and natural sounds commonly occurring
in suburban areas. Based on the results, a linear regression model was
proposed, which explains well the overall pleasantness evaluation of two
combined stationary sounds based on the weighted sum of the singular ratings
and their interaction. In this model, unpleasant sounds receive a greater
weight compared to pleasant ones, which presumably is due to negativity
dominance and partial masking effects. A second listening study was
performed with the same test design as in the first experiment, but with
three combined sounds. The results from the second experiment were used to
validate the existing regression model. An extended model predicting
overall pleasantness evaluations based on singular ratings helps
establishing a perception-oriented classification of complex acoustical
environments and the calculation of noise surcharges.