Development of an evaluation method of acoustic vehicle alerting system

Abstract

Using multiple linear regression analysis, we defined equations to describe basic psycho-acoustic quantities, such as sharpness, tonality, roughness and fluctuation strength, to estimate how people react to various types of auditory signals. The HEV/EV auditory signal is a very important audio-guidance for pedestrians. It must be recognizable in a city crowd, as well as within a comfortable hearing range. In common practice, sensory evaluation tests are used to estimate the human auditory experience, but these tests are expensive due to the large size of the required test pool. Through investigating the correlation between sensory evaluation test results and basic psycho-acoustic quantities for 22 signals, we developed equations that predict the results of sensory evaluation tests inexpensively. Our method has been confirmed through the agreement between results obtained from our equations and that of sensory evaluations. We showed that the tendency of actual sensory evaluation could be predicted by the prediction relation of sensory evaluation using the metrics of sharpness, tonality, roughness, and fluctuation strength. With the increase in the number of samples, the dispersion in the prediction range spread. Therefore, it seems that the related item which is not taken into the present consideration needs to be added for improvement in the further predictive accuracy.