Auralization of simulated tyre noise: Psychoacoustic validation of a combined model

Artikel i vetenskaplig tidskrift, 2019

Due to improvements on combustion-engines and electric-engines for cars, tyre noise has become the prominent noise source at low and medium speeds. Models exist that simulate the noise produced by a rolling tyre, as do models that auralize different traffic situations from a basic data set. In this paper, an established model for tyre noise (SPERoN) is combined with an auralization tool. The combined model can predict the spectrum of the sound at 7.5 m, as well as reproduce the sound for a given listener position. The auralization uses a methodology where recorded sounds are converted to source signals for engine and tyre/road-interaction. These can be shaped by the spectra estimated in SPERoN and synthesized back into a pass-by signal. Psychoacoustic judgements were used to compare the modelled signals with recorded signals. To see how well the modelled signals match the real recorded signals for perception, two listening-tests were performed. The simulated and recorded signals were rated by pleasantness, loudness, roughness and sharpness using semantic differentials. It was found that responses for simulated and recorded signals correlate for all cases, but rankings could not be reproduced exactly. The model can be further improved to be more applicable for listening tests.