The Auditory System at the Cocktail Party

Abstract

Most normal-hearing listeners can understand a conversational partner in an everyday setting with an ease that is unmatched by any computational algorithm available today. This ability to reliably extract meaning from a sound source in a mixture of competing sources relies on the fact that natural, meaningful sounds have structure in both time and frequency. Such structure supports two processes that enable humans and animals to solve the cocktail party problem: auditory object formation and auditory object selection. These processes, which are closely inter- twined and difficult to isolate, are linked to previous work on auditory scene analysis and auditory attention, respectively. This chapter considers how the brain may implement object formation and object selection. Specifically, the chapter focuses on how different regions of the brain cooperate to isolate the neural rep- resentation of sound coming from a source of interest and enhance it while sup- pressing the responses to distracting or unimportant sounds in a sound mixture.