The Spectral Structure of Acoustic Time Series Can Predict the Perceptual Assessment of Urban Soundscapes

Andrew Mitchell,Jian Kang

DOI: https://doi.org/10.1121/1.5136681

2019-01-01

The Journal of the Acoustical Society of America

Abstract:The field of soundscape studies considers sound environments as perceived, in context, and has recently focussed on understanding the potential of physical acoustical features, including temporal characteristics, for predicting human perception. This study investigates the presence of a 1/f structure in the power spectrum slope of six (psycho)acoustical parameters of 30-s recordings of urban acoustic environments. The acoustical parameters were calculated as time series throughout the recording period, then the power spectrum of the time series was calculated and plotted on a log-log scale, with the x-axis ranging from 30 s/10-1.5 Hz to 0.01 s/102 Hz. The slope of the best-fitted straight line through the power spectrum was calculated and compared to the corresponding perceptual attribute ratings of the soundscape collected on site during the recording. The dataset includes 300 + recording-response pairs. An overall 1/f structure was not found for any of the parameters. Differences in temporal behaviour are indicated at different time scales, with a deviation in slope typically occurring at ∼2 s/0.5 Hz, reflecting differences in temporal behaviour among within-sound-event time scales and between-sound-event time scales. An ordinal logistic regression model is developed which predicts the perceptual attributes of urban soundscapes based on the spectral slopes of the acoustic time series.