Environmental noise caused by traffic and industrial facilities is a relevant health factor in urban areas, along major traffic routes and in the vicinity of airports. The technique of auralization helps exploring the relationships between sound and its impact on people under fully controlled conditions. Further, by directly addressing the hearing sensation, it is an intuitive tool for the assessment and communication of environmental noise scenarios. Analogous to visualization, auralization enables us to listen to situations that do not exist in reality.
In this thesis research, methods for the auralization of environmental acoustical sceneries are established. The sceneries are represented by a virtual environment containing virtual sound sources that are arranged in space and time, and within which sound waves propagate to a virtual observer. To that aim, sophisticated calculation models for the synthesis and reproduction of road traffic, railway and wind turbine noise are developed. This requires investigating the relevance of the involved acoustical phenomena for perceived realism. On that basis, calculation models (i.e. synthesizer structures) are proposed that adequately reproduce source characteristics, sound propagation effects, and spatial impression. The models are accompanied by methods to derive the necessary input parameters from own measurements and data analysis.
The presented calculation models are parametric and thus allow for a large versatility with respect to scenarios and sound reproduction. Because the three considered environmental noise sources feature their specific acoustical peculiarities, source-specific models are proposed. These source-specific models have in common that the sound radiated by a source is artificially generated using digital sound synthesis. For wind turbine and road traffic noise, a combination of additive and subtractive synthesis, denoted as spectral modeling synthesis, is applied. A uniqueness of the wind turbine synthesizer is the ability to reproduce and control different types of characteristic amplitude modulation. The synthesizer for road vehicles separately produces tire noise and propulsion noise. The generated propulsion sounds depend on the engine type, the instantaneous engine condition (engine speed and load), and the emission angle. An additional special feature of the propulsion sound synthesis is the fact that, besides amplitude and frequency, the phase of the engine harmonics has to and is considered.
For railway rolling and impact noise, in contrast, a physically-based synthesis approach has been developed that describes the mechanical excitation and the vibration of the dynamic wheel/rail system. The corresponding model considers the microstructure of the wheels and rails, as well as structural resonances of the wheel/rail system to elicit the typical metallic sound character of railway noise. In all models, sound propagation effects, such as geometrical divergence, Doppler effect, atmospheric absorption, ground effect and amplitude fluctuations due to atmospheric turbulence, from a virtual point source to a virtual observer location are simulated by processing the synthetic source signals with time-variant filters in the time domain.
Auralizations created with the presented models feature a high audio quality and are judged as plausible and realistic by expert listeners. To achieve this realism in the auralizations, it was found that variation with respect to time, frequency, space, and orientation is crucial.
The presented models extend the today's body of existing auralization models and allow for new possible applications.