Auralization of electric vehicles for the perceptual evaluation of acoustic vehicle alerting systems

Abstract

Electric vehicles (EVs) typically emit little noise at low driving speeds, which increases the risk of accidents for vulnerable road users such as pedestrians. To reduce this risk, regulations demand that newly sold EVs have to be equipped with an acoustic vehicle alerting system (AVAS), which radiates artificial warning sounds. Developing AVAS sounds that provide a sufficient warning capability while limiting traffic noise annoyance requires laboratory listening experiments; such experiments need accurate auralization methods. Even though several auralization tools are already established in the research field, those frameworks require additional data to simulate EVs. This paper presents an electric vehicle auralization toolchain combined with an open-access database, including AVAS measurements, synthesis algorithms, and numerically calculated sound source directivities for three different electric passenger cars. The auralization method was validated numerically and in a listening experiment, comparing simulated EV passages to binaural in-situ recordings. The results of this perceptual validation indicate that stimuli generated with the presented method are perceived as slightly less plausible than in-situ recordings and that they result in a similar distribution of annoyance ratings but a higher perceived vehicle velocity compared to the reference recordings.