Characterizing the Familiar-Voice Benefit to Intelligibility
Abstract
Everyday listening often occurs in the presence of background noise. Listeners with normal hearing can often successfully segregate competing sounds from the signal of interest. To do this, listeners exploit a variety of cues to facilitate the separation of simultaneous sounds into separate sources, and group sequential sounds into intelligible speech streams. One of the cues that has been shown to be an effective facilitator of speech intelligibility is familiarity with a talker’s voice. A recent study by Johnsrude et al. (2013) measured speech intelligibility of a naturally familiar voice (i.e., that of a long-term spouse) and showed a large improvement in intelligibility when a spouse’s voice serves as the target or the masker. This improvement is commensurate with another cue that is well-understood to be a strong facilitator of intelligibility: spatially separating two speech streams. Therefore, the goal of this thesis is to extend the work of Johnsrude et al. (2013) by providing a clearer understanding of voice familiarity as a cue for improving intelligibility. Specifically, the aims of this thesis are 1) to measure the magnitude of intelligibility benefit of different types of naturally familiar voices: friends and spouses, (2) to quantify the familiar-voice benefit in terms of degrees of spatial separation, and (3) to compare the neural bases of voice familiarity and spatial release from masking to determine if these cues improve intelligibility by recruiting similar areas of the brain. The primary findings of this thesis were that 1) the familiar-voice benefit of friends and spouses are comparable to each other and that relationship duration does not affect the magnitude of the familiar-voice benefit, (2) that participants gain a similar benefit from a familiar target as when an unfamiliar voice is separated from two symmetrical maskers by approximately 15° azimuth, and (3) that familiar voices and spatial release from masking both activate known temporal voice areas, but attending to an unfamiliar target voice when masked by a familiar voice also recruits attention areas. Taken together, this thesis illustrates the effectiveness of a naturally familiar target voice in improving intelligibility.