Railway psychoacoustic annoyance - an indicator for planning infrastructure projects

The joint project "Real-time Calculation, Auralization and Visualization of Sound Propagation and Noise Protection Measures for Infrastructure Projects" (EAV-Infra), funded by the German Federal Ministry for Economic Affairs, investigates the potentials of digital infrastructure planning based on Building Information Modeling (BIM) in combination with state-of-the art Auralization and Visualization technology as well as Psychoacoustic Modelling. Currently, log-scaled sound pressure levels, which are only informative to experts, are used to communicate the topic of noise exposure to the public. Auralization, in combination with a hearing-oriented noise index, will constitute a more intuitive and understandable evaluation basis for residents. One focal point of EAV-Infra is the development of a linearly-scaled metric for railway noise, which correlates better with human perception than the LAeq. Binaural recordings of a wide spectrum of different railway pass-by scenarios were presented in listening experiments to obtain semantic-differential information about the sounds as well as global and continuous annoyance ratings for individual passings. Currently, a railway psychoacoustic annoyance metric is being developed by comparing two fitting methods: a classical model based on common psychoacoustic quantities, as well as an AI model with a Mel-log-spectrogram-CRNN architecture. The approach and first results will be discussed in this contribution.