Validation of Different Synthesis Methods of Structure Borne Road Noise in a Virtual Car Sound Environment

摘要

The pressure on the development cycles in automotive industry forces acoustic engineers to create awareness of in-vehicle acoustic comfort in the earliest stages of development. Over the last few years, significant research efforts have been done to develop compact Sound Quality Equivalent models for engine noise, road noise and wind noise and to implement them in an interactive, real-time Virtual Car Sound environment. The research presented in this paper reports on four different synthesis methods that have been developed for structure borne road noise. The first method starts from white noise that is filtered in 1/3th octave bands and shaped with velocity dependent levels. The second approach is similar, but it makes use of one global velocity scaling factor for every frequency band. The third method is a time-randomisation approach that starts from a reference spectrum, while the last one is based upon the integration of time structures. A number of experiments was carried out on different roads (soft Asphalt, rough Asphalt, concrete road with seals) and under different conditions (constant speed, time-varying speed). Jury tests were organised and psycho-acoustic metrics (loudness, roughness, fluctuation strength,…) were calculated to evaluate and compare the road noise synthesis methods.