It should be possible to carry out dynamic range compression in hearing devices, and in particular in hearing aids, so it is free from distortion and practically in real time. For this purpose it is proposed that the modulation spectrum be obtained from the audio or input signal. The modulation spectrum is subsequently directly modified corresponding to a predefined compression function. Finally a modified or compressed output signal is recovered from the modified modulation spectrum. Alternatively a complex envelope may be obtained from the input signal, which is filtered using time-variant modulation filtering corresponding to a predefined compression rule. A distortion-free, compressed output signal may also be recovered herefrom.