Abstract:   In a noisy environment with multiple-people talking, how does a listener use certain perceptual cues to facilitate recognition of target speech? To answer this “cocktail-party” problem, researchers have used brain-image methods to examine the underlying brain-network mechanisms. Recent studies have shown that using a unmasking cue, which is associated with a feature of target speech, is able to not only shorten the latency of the auditory cortical responses to a target-speech signal, but also facilitate activation and functional connection of the brain regions that are associated with both cue-specific and cue-nonspecific attention, speech expression, inhibitory functions, and speech motoring, thereby forming the brain-network foundation for the unmasking-cue-induced improvement of the target speech perception against speech informational masking.

Key words: cocktail-party problem, informational masking, unmasking, speech recognition, perceived spatial separation, auditory speech priming, visual speech priming, brain network, neuroimaging