Apparatus configured to predict a diaphragm deflection signal, block-by-block, in overlapping time blocks based on the loudspeaker signal to obtain one diaphragm deflection signal block per time block. The apparatus is configured to determine a temporal position of a maximum deflection of a current diaphragm deflection signal block of a current time block within an overlap area with a subsequent time block and to calculate a level up to which the current diaphragm deflection signal block can be controlled without diaphragm over-deflection for the current time block by considering a comparison of the current diaphragm deflection signal block with a subsequent diaphragm deflection signal block or an estimation of the subsequent diaphragm deflection signal block from the current diaphragm deflection signal block at the temporal position. The apparatus is configured to attenuate the current diaphragm deflection signal block and to synthesize a modified loudspeaker signal.