The University of Illinois (UIUC) is working with Apple and other technology giants on the Speech Accessibility Project, which aims to improve current versions of difficult-to-understand speech recognition systems for people with speech patterns and disabilities. Although often derided for mishearing users’ requests, the voice recognition systems of digital assistants like Siri have become more accurate over the years, including developing recognition on devices. In a new initiative, a project aims to further improve accuracy by targeting people with speech impairments and disabilities.
Working with Apple, Amazon, Google, Meta and Microsoft, as well as nonprofit organizations, UIUC’s Speech Accessibility Project will attempt to expand the range of speech patterns that speech recognition systems can understand. This includes a focus on speech affected by disease and disability, including Lou Gehrig’s disease, amyotrophic lateral sclerosis, Parkinson’s disease, cerebral palsy and Down syndrome.
In some cases, speech recognition systems can provide quality-of-life improvements for users with disorders that inhibit movement, but issues affecting the user’s voice can affect their effectiveness.
Under the Speech Accessibility Project, researchers will collect samples from individuals “representing different speech patterns” to create a private and non-identified dataset. This dataset will initially focus on American English and can then be used to train machine learning models to better respond to speech.
The participation of a range of technology companies that have virtual assistants or offer speech recognition in their tools could help accelerate the project. Instead of using separate teams that might duplicate results found by others, these teams could instead collaborate directly through the project.
“The speech interface should be open to everyone, and that includes people with disabilities,” says Mark Hasegawa-Johnson, a professor at UIUC.” This task has been difficult because it requires significant infrastructure, preferably the kind that can be supported by leading technology companies, so we created a unique interdisciplinary team with expertise in linguistics, speech, artificial intelligence, security and privacy.”
Visit the project’s official website to learn more at
