The Urban Acoustic Environment as a Complex Network

Abstract

The urban acoustic environment (AE) contains valuable information on complex sub-systems of urban areas, such as traffic or biodiversity. As the availability of cheap sensors and computational power increases, so does the need for methods to process high-dimensional audio data. We take this as an opportunity to introduce complex networks (CNs) to the field of urban acoustics. CNs have proved effective in capturing the complexity of e.g. climate dynamics or brain structures, and thus, represent a promising tool for the high-dimensional urban AE. We present how CNs are constructed based on frequency correlation matrices and show their behavior for various time periods and sound sources. To demonstrate their use on audio data from the urban environment, we apply them to the dataset from the SALVE study to systematically characterize the urban AE. Here, we use subsets of SALVE with two different temporal resolutions: 1 s over three minutes and three minutes over 24 h. Measures such as the average shortest path length identify urban sites with similar AEs, indicating the utility of CNs to identify non-random patterns in large datasets of the urban AE.