In this paper, a hybrid computational aero/hydro-acoustic approach is proposed to deal with acoustic scattering and flow-induced noise problems based on the sharp interface immersed boundary method (IBM). For the flow field, the incompressible Navier-Stokes equations are solved by an in-house direct numerical simulation solver. The acoustic field is predicted by solving acoustic perturbation equations (APEs). Both flow and acoustic solid boundaries with complexity and mobility are dealt with by the sharp interface IBM. Benchmark acoustic problems with varied scatterers in two and three dimensions are presented to validate the accuracy of the acoustic codes and boundary treatments. Then, the feasibility and accuracy of the present hybrid approach are validated by considering the problem of flow past a circular cylinder at a Reynolds number of 200. Subsequently, the present method is used to predict the noise generated by flow around a four-cylinder array in two-dimensions with two arrangements (i.e., square array and diamond array), and the flow and acoustic physics are investigated in detail. The results show that the square array retains a monopole-like sound-radiation shape, while the directivity pattern of the diamond array produces a dipole-like shape. In both the square and diamond arrays, the propagation of acoustic waves is affected by the Doppler effect, and the latter array results in a larger alternation of the propagation angle compared with the single cylinder due to the influence of the geometric configuration. The intensity of the radiated acoustic pressure is much greater for the diamond array compared to the square one in most circumferential directions, and the acoustic intensity of both arrays is greater than that of the single cylinder. The spectrums of the far-field acoustic pressure indicate that the two arrays and the single cylinder have similar peak frequencies and profiles, with vortex shedding playing the predominant role in noise generation in all three configurations. (c) 2021 Published by Elsevier Ltd.