Exploring auralizations with multiple doppler effects for urban sound planning

Abstract

Increasing noise levels due to urban densification is a major concern in many countries. Long-term noise exposure is associated with a long list of health risks, including heart attacks, diabetes, and sleep disorders, to mention a few. To mitigate such health risks, thorough urban sound planning is a must. Noise prediction models are often used to estimate A-weighted sound pressure levels in a sound planning process. Nevertheless, solitary use of sound pressure levels is not sufficient to evaluate the psychoacoustic aspect of sound. Auralizations were therefore proposed in this thesis as a complement to existing noise prediction models in urban sound planning. Moreover, a specific aim was to validate the plausibility of auralizations with multiple Doppler effects versus a single Doppler effect for pass-by vehicles. This objective was tested in different virtual urban scenes, with and without edge diffraction. A listening experiment was conducted for this purpose, and in the first part, a paired comparison between a single Doppler and multiple Doppler stimuli, the majority of the participants voted for the multiple Doppler stimuli as more plausible. A two-sided Fisher’s exact test on these results gave a resultant p-value of 0.0004, indicating a significant difference in preference between single and multiple Doppler effects. Furthermore, an odds ratio of 0.247 indicated that the odds of preferring the multiple Doppler stimuli were higher than for the single Doppler stimuli. The second part of the listening experiment intended to find an appropriate sample rate of the source position. The results showed that more complex urban scenes, e.g., with diffraction from a noise screen, required higher sampling rates to achieve more plausible auralizations with smoother transitions to shadow zones. Source position sample rates down to 10 Hz seemed sufficient in simple scenes. Ambient sound was recorded and added to a set of pass-by auralizations in the last part of the listening test. A paired comparison between stimuli with and without ambient sound was carried out to check if the ambient sound made an improvement to the perceived plausibility. Increased plausibility from ambient sound could not be proved due to the small sample size.