Grant: Discovery Grants Program - Individual

Following and understanding a conversational partner, despite interference from other sources, is a feat most of us accomplish every day. However, the ability to identify, track, and understand a voice, particularly when other sources of sound (including other voices and their echoes) are simultaneously present, is nontrivial. Even sophisticated artificial-intelligence systems cannot yet achieve the levels of performance that people routinely exhibit. The human ability to segregate one sound source from a 3-dimensional auditory "scene" depends on differences in acoustic features between that sound and others in the mixture. In addition to low-level, acoustic, cues, prior knowledge, such as familiarity with the talker's voice, may influence auditory scene analysis in important ways, but this is almost entirely unexplored. Auditory scene analysis (ASA) refers to the processes by which a complex and dynamic pattern of vibration in the ear (a mixture of all the sounds in the environment) is resolved by the brain into individual sound sources in different locations, so that one (e.g., a voice) can be tracked and understood. In the next five years, we will extend our previous work investigating how knowledge of a familiar voice, such as that of a spouse, sibling or friend, enhances intelligibility. Previous work was headphone-based and so neglected spatial listening; now we will examine familiar voice benefits in more naturalistic conditions using our new virtual acoustic space (the "AudioDome"), a state-of-the-art 95-loudspeaker array capable of 9th order ambisonic rendering. The proposed research has 3 aims: 1) to extend our work to more naturalistic, 3-dimensional auditory scenes and examine whether familiar voices facilitate auditory scene analysis; 2) to identify the voice characteristics that are most important for enabling the intelligibility benefit from familiar voices; and 3) to better understand how the cognitive and neural processes recruited during listening differ for familiar and unfamiliar voice targets.  Our work uses psychophysical behavioural, as well as EEG and pupillometric methods, both in the AudioDome and in sound-attenuating booths. in most of our experiments we will test younger (18-35 years old) listeners. In a few experiments we will also test older (55 - 80 years) normally hearing and clinically hearing impaired participants; the latter with well-fitted hearing aids. Through the proposed research I will train several students and 2 postdoctoral fellows.  The overall goal of our research is to develop a more detailed account of how knowledge contributes to efficient auditory perception, including ASA. The central role of prior knowledge in hearing must be recognized if we are to effectively tackle the problem of impaired hearing, enabling older people to take more complete advantage of their experience, and ability to learn, in order to mitigate challenges in everyday, multisource, listening environments.