Our work aims to further our understanding of the normal processes involved in the perception of speech. We examine the acoustic cues used by the normal auditory system and extent to which these cues are available to the impaired or implanted auditory system.

These cues involve frequency resolution, temporal resolution, and the integration of these cues.

Basic Understanding of Speech in Noise

Related work involves the perception of speech in background noise. The normal human auditory system has a remarkable ability to extract speech from noise. But this skill is often lost in cases of hearing loss. In fact, poor speech understanding when noise is present is the number one complaint of people with hearing loss, and so this is one of our most important problems.

We examine the role of various acoustic cues used to extract speech from noise, and how these cues are disrupted when hearing loss is present.

Tools to Improve Speech in Noise

We are also involved in the development of tools aims to alleviate the speech-in-noise limitations of hearing loss. In one line of work, techniques of deep learning are employed to identify regions of speech and regions of noise, and remove the latter. This results in substantially increased speech understanding. It is our hope and expectation that these technologies can be implemented into future hearing aids, cochlear implants, and other auditory technologies.