Background traffic noise synthesis

Paper in proceeding, 2016

When planning the development of urban areas, it is important to assess the future acoustic environment. Currently, this evaluation is achieved with the help of acoustic indicators, but they do not suffice for a holistic description of the perceived sound environment. New indicators can be extracted through listening tests and analysis of different acoustic scenarios. However, generating such scenarios using auralisation models for outdoors sound propagation is often computationally highly demanding. Here, a simplified auralisation model is described, focusing on background traffic noise simulation on flat city scenarios. For computational efficiency, the proposed method partly relies on physical models for air attenuation, ground effects and spherical spreading. The doppler effect and the contribution of individual vehicle pass-bys are achieved with the help of modulation transfer functions, and spatial imagery is realised by both non-corellated phase spectra and modulation transfer functions. Power profiles from measurements are used to model rolling noise. The proposed model is assessed through listening tests against the LISTEN demonstrator on its perceived speed and distance from the listener. The perceived speed is matching better to the LISTEN between 70 kmph and 90 kmph, while above 300 m and up to 900 m from the source, the distance is more correctly guessed from the subjects.