Energetic Masking and Masking Release

Abstract

Masking is of central interest in the cocktail party problem, because interfering voices may be sufficiently intense or numerous to mask the voice to which the listener is attending, rendering its discourse unintelligible. The definition of energetic masking is problematic, but it may be considered to consist of effects by which an interfering sound disrupts the processing of the speech signal in the lower levels of the auditory system. Maskers can affect speech intelligibility by overwhelming its representation on the auditory nerve and by obscuring its amplitude modulations. A release from energetic masking is obtained by using mechanisms at these lower levels that can recover a useful representation of the speech. These mechanisms can exploit differences between the target and masking speech such as in harmonic structure or in interaural time delay. They can also exploit short-term dips in masker strength or improvements in speech-to-masker ratio at one or other ear. ?