A method for processing audio signals in particular for a therapy of subjective tinnitus with an individual tinnitus frequency. The method includes: providing a first audio signal, determining a blocking range in the frequency spectrum of the first audio signal with a predefinable frequency width, creating a second audio signal from the first audio signal using a filter, and determining an auditory energy of the first audio signal or the second audio signal within at least one predefined therapeutically applicable frequency range and specifying an evaluation parameter for the second audio signal.