Modelo para la predicción de la estructura temporal y espectral de los paisajes sonoros urbanos basado en redes neuronales

Abstract

The urban soundscapes are dominated by the noise generated by the road traffic, which acts as a noisy source that masks the rest of present sounds, which creates diverse problems that go beyond that the annoyance and displeasure in the population. On the other hand, the perception of the soundscapes by the population is not explained only for the sound pressure level to which it is exposed, but temporal and spectral characteristics of the soundscape must be taking into account, as well as other variables not acoustics, e.g., the attitude towards the source, sensitivity and habituation to the noise, meant to the population, etc. All this supposes that, as different authors establish and, within the framework of the current European legislation, the design, planning and assessment of the soundscapes should be included in the future town planning. In this paper a model, based on neural networks, for the prediction of the temporal and spectral structure of the soundscapes is proposed, which, together with the perceptual assessment of the soundscapes, allows us the adjustment of these to the needs and requirements of the population in the urban environments. Likewise, a comparison between the predicted and the experimentally obtained data in diverse sound scenes of the city of Granada is presented.