“What” perception of sounds and words
Melissa M. Baese‐Berk · The Journal of the Acoustical Society of America · 2019
Speech perception is a task that, on the surface, is deceptively simple. Every day in most of our interactions, we are able to understand speech from a variety of talkers about a variety of topics without significant challenges. However, the actual task of understanding speech is a complex one, requiring a listener to quickly extract information from a noisy and variable acoustic signal to determine the speaker’s intended meaning. Determining what linguistic features and words a speaker has intended is not trivial, as many factors can influence the acoustic signal. That is, the same intended word or speech segment can be produced with many distinct acoustic realizations depending on the speaker, the surrounding speech context, and other linguistic features, including the speaking rate. Further, speech perception can be even more challenging when a listener is trying to perceive speech in noisy environments, including environmental noise (e.g., construction noise) or competing speech from other talkers. In this talk, I will focus on how listeners are able to extract relatively invariant meaning representations from a highly variable signal. Understanding what is perceived when listeners hear speech is critically important for an understanding of what factors matter for speech privacy.