Skip to content. Skip to navigation
CIM Menus

Tongues, brains, robots, and Cinderella: Different ways to assess atypical speech

Frank Rudzicz
Thotra Inc.

August 11, 2015 at  11:00 AM
McConnell Engineering Room 437


In this talk I will survey a few of our projects that relate to assessing various types of speech disorders. We start with a look at physical speech disorders caused, for example, by cerebral palsy. For this, we use a technique called electromagnetic articulography to get fine measurements of the vocal tract during speech and relate these to the resulting acoustics using differential models of physical speech production. A lot of the machine learning approaches used in that work transfer to measuring cognitive speech disorders caused, for example, by Alzheimer’s disease. There, we measure acoustic and lexical/syntactic aspects of the voice to assess dementia in patients from small snippets of speech as they talk, e.g., about Cinderella. Similar measures can also be used on data from the brain, via electroencephalography, to learn and identify semantic and phonological differences there. Finally, in order to elicit speech from individuals with dementia, we are exploring the use of a small humanoid robot to engage in conversation, which will involve identifying periods of confusion in human speakers.


Dr. Rudzicz is the author of about 60 papers generally about natural language processing but focusing mostly on atypical speech and language in individuals with physical disorders (e.g., cerebral palsy, Parkinson’s disease) and in individuals with cognitive disorders (e.g., Alzheimer’s disease). He is the founder and CEO of a company, Thotra Inc., that transforms speech signals to be more intelligible, a co-author on a best student paper award at Interspeech 2013, an Ontario Brain Institute entrepreneur, and winner of the Alzheimer’s Society Young Investigator award. Dr. Rudzicz is a guest editor for special issues of the ACM Transactions on Accessible Computing, and Computer Speech and Language, and the President of the joint ACL-ISCA Special Interest Group on Speech and Language Processing for Assistive Technologies (SLPAT). He developed the speech recognition and interaction being used in HitchBOT, the hitchhiking robot, which has received international attention.