Irritating earthquake sounds, reported also at low ground shaking levels, can negatively impact the social acceptance of geo-engineering applications. Concurringly, earthquake sound patterns have been linked to faulting mechanisms, thus opening possibilities for earthquake source characterisation. Inspired by consistent reports of felt and heard disturbances associated with the weeks-long stimulation of a 6 kmdeep geothermal system in 2018 below the Otaniemi district of Espoo, Helsinki, we conduct fully-coupled 3D numerical simulations of wave propagation in solid Earth and the atmosphere. We assess the sensitivity of ground shaking and audible noise distributions to the source geometry of small induced earthquakes, using the largest recorded event in 2018 of magnitude M L =1.8. Utilizing recent computational advances, we are able to model seismo-acoustic frequencies up to 25 Hz therefore reaching the lower limit of human sound sensitivity. Our results provide for the first time synthetic spatial nuisance distributions of audible sounds at the 50 -100 m scale for a densely populated metropolitan region. In five here presented 3D coupled elastic-acoustic scenario simulations, we include the effects of topography and subsurface structure, and analyse the audible loudness of earthquake generated acoustic waves. We can show that in our region of interest, S-waves are generating the loudest sound disturbance. We compare our sound nuisance distributions to commonly used empirical relationships using statistical analysis. We find that our 3D synthetics are generally smaller than predicted empirically, and that the interplay of source-mechanism specific radiation pattern and topography can add considerable non-linear contributions. Our study highlights the complexity and information content of spatially variable audible effects, even when caused by very small earthquakes.