The UA-Speech corpus was published in 2008. It originally contained recordings of nineteen individuals with dysarthria as a correlate of cerebral palsy, plus age-matched and gender-matched controls. Of the original nineteen, three have removed permission over the years, and/or their data has been corrupted, so that the distribution now contains data from sixteen speakers. Information about the corpus is available in our paper, Dysarthric speech database for universal access research.

Collection of the UA Speech dataset was approved by the University of Illinois Institutional Review Board, project titled “Audiovisual Description and Recognition of Audible and Visible Dysarthric Phonology,” funded as 5R21 DC008090 by the NIH NIDCD, 2006-2009.

The original participant consent forms specify the following constraints: Data are available only to researchers at government and academic research labs. Commercial uses are not permitted. Data must be protected against theft, and may not be redistributed. The full license is available in the distribution, in the file LICENSE.txt.

If you are interested in obtaining a copy of the dataset, send e-mail to In your e-mail, please specify:

  • Your name
  • Your institution name (school or government institution)
  • Intended use of the corpus
  • Your institutional e-mail address
  • An account that you have with google, icloud, or onedrive that can be used to deliver the data.
  • The period of time for which you will need access to the data.