A long-term goal of our research is to identify subtle speech deficits in people with neurodegenerative diseases such as Parkinson disease (PD) in order to improve their speech outcomes and quality of life, through early detection, and continuous monitoring of deficits. One of the primary confounding factors that currently limits early identification is the utilization of simple elicitation tasks (e.g., sustained vowel phonation, reading), where speakers are capable of compensating for small perturbations to the speech system. Our approach to overcome this limitation is to design tasks that will impose sufficient burden on speakers’ speech motor control in order to characterize the use of any compensatory behaviors. In this domain, our research will focus on the development, characterization, and validation of a) novel complex vocal motor tasks, and b) dual tasks.
Our lab is interested in the development of novel, automated acoustic measures (e.g., cepstral measures of articulation) that describe and differentiate neurodegenerative diseases such as PD. These sensitive measures of speech production can be used to track changes in the disease progression, and can be used to evaluate clinical progress. In addition, our lab will also examine acoustic–perceptual relationships of ‘speech naturalness’ using novel acoustic measures of multiple prosodic dimensions (e.g., monopitch). Speech naturalness translates to an individual's participation at the societal level. Thus, from a rehabilitation perspective, exploring speech naturalness and communicative participation using both perceptual and acoustic measures of prosody is an important clinical endeavor.
Our lab actively collaborates with the Auditory and Speech Sciences Lab (ASSL) directed by Dr. David Eddins, on several voice quality projects. Our research in this domain focuses on precise assessment of voice quality changes through the development of new psychometric scales. Auditory perceptual evaluation of dysphonic voice quality is one of the most common and valuable clinical tools for determining severity of voice disorders and measuring treatment outcomes. Terminologies such as “breathiness”, “roughness”, “strain” are often used to represent perceptions of dysphonic voice quality, and are widely used in standardized voice assessment such as Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V). Understanding the acoustic, perceptual, and physiological correlates of these voice quality percepts is an essential component of voice research and clinical practice, especially because of their relationships to physical and neurological pathologies (e.g., spasmodic dysphonia, PD). A long term goal of this work, is to develop automated estimates of voice quality based on computational models of auditory perceptual behaviors.