Sound and the city: Current perspectives on acoustic geo-sensing in urban environment

Acoustic geo-sensing deals with the derivation of spatial information from audio data. Motivated by the rising field of sensing information in (smart) cities, we present current perspectives and a methodological update on the acoustic-based solution of recognition problems in a complex urban real-world scenario. In particular, we examine a static and a sequential approach for the automatic recognition of a cyclist’s route between fixed endpoints, the route direction, and the route progress, solely based on the audio data recorded with a cell-phone attached to the cyclist. To this end, we introduce the freely available ‘Graz Cell-phone Cycle Corpus extended’ (GC₊ ³ ) that consists of more than 25 h of GPS-annotated audio material recorded in the southern urban area of the city of Graz, Austria. Promising recognition results confirm basic feasibility of acoustic geo-sensing in an urban environment. The approach’s theoretical potential of being generalised to support the solution of position tracking problems on a multi-dimensional urban grid gives reason for the implementation of acoustic geo-sensing methodology in a number of real-world applications.