The goal of this project was to develop a computer program for real-time acoustic simulation of the noise in an operating truck. The approach was twofold. Firstly, sounds and noise were recorded via a microphone in a truck, running under various conditions in the real world (different driving speeds, engine revolutions while driving or during a standstill, the wind and wheel noise of the moving truck without working engine). The recorded noises were analysed using Fast Fourier Transformation to get the power spectrum in steps of third octaves. From these data the relevant elements were extracted constituting the vehicle noise under various conditions. Secondly, a noise model was developed based on the relevant physical parameters of a truck in action. The empirical findings and the model with its parameters were combined and programmed to run on the sound machine of a Silicon Graphics workstation. The sound impression of different developmental versions of the program were evaluated with respect to the noise impression in reality. The result of the project is that the physically perfect noise model did not produce the best impression. Instead, adding some random noise caused a much more realistic impression. (A)
Abstract