Electric motors have become a part of our daily lives, making the question of their noise essential for our acoustic comfort. The sound they emit is often described as tonal or whistling due to the presence of many harmonics, one source of which is a power supply technique called pulse width modulation (PWM). During the design process, auralization models can be used to evaluate the effect of PWM harmonics on sound quality. Engineering-grade models, which are based on simplifying assumptions, are used in the early stages of design when little input data is available. With these models, a direct perceptual comparison between measured and simulated sounds would inevitably reveal significant differences. However, if the model can reliably predict the perceptual sound space, it can be a valuable tool for assessing sound quality. This paper presents a methodology to evaluate the capability of a simplified engineering model to simulate the main attributes of PWM noise. To this end, the authors implemented an auralization model to synthesize the noise emitted by an electric motor from its supply signals. Some sound stimuli were collected from measurements and simulations to conduct a perceptual experiment. The measured and simulated sounds were evaluated separately within two sets of stimuli. The experiment included similarity and pleasantness evaluations. Comparing the results obtained by Individual Difference Scaling (INDSCAL) showed great coherence between the two sound sets, suggesting that the simulated stimuli were evaluated similarly to the measured stimuli. Pleasantness ratings yielded the same result. Therefore, the auralization model appears to reliably reproduce the main sound dimensions underlying the perception of PWM noise. • An auralization model for e-motor noise is validated through a perceptual experiment. • INDSCAL analysis provides similar 2D perceptual spaces for real and virtual stimuli. • Low-frequency noise and rattle noise affect the perception of electric motor noise. • The similarity of pleasantness ratings demonstrates the relevance of the auralization model.
Wanty et al. (Wed,) studied this question.