Identifying sound sources in terms of urban environmental parameters

It is well known that an urban environment is essential to determine its surrounding sounds. This paper explores relationships between urban environmental parameters and the embracing sound sources by regression models, and then identifying the sound sources according to diversity urban environmental parameters. Previous studies of sound identification paid more attentions on acoustic parameters variety but ignored urban environment differences. In this study, large number of sound sources have been collected through intensive quantity field studies in Shenzhen, which is a coastal metropolis in southern China having typical ecological urban environments. Firstly, the sound sources in collected recordings were identified by listening and then the proportion of sound sources including natural, living, mechanical sound was counted. Based on SPSS analysis of physical and psychological features of a sound source and its linked environment, regression models have been developed to study their relationships. Afterwards, sound sources can be identified using the developed regression models. The results find that the relationships between sound sources and urban environmental parameters including FAR, Building Coverage, Road Density and Greening Ratio are all significant. The models of identifying natural and mechanical sound sources performed better. But living sound sources of recordings could not be identified by liner regression models easily.