The aim of the invention is to achieve a cleaner separation of a first audio signal in a first region of an area to be exposed to sound emitted by a plurality of loudspeakers. For this purpose, a calculating element calculates the version of the audio signals resulting from the spatially selective playback of the audio signals in this first region, calculates a masking threshold on the basis of the version of the audio signal that is to be separated from the one or more other audio signals in this region, and influences the output of the audio signals for the spatially selective playback to the outputs of the plurality of loudspeakers on the basis of a comparison of the masking threshold with the version of one or more other, i.e. interfering, audio signals.