Tech Giants Launch Speech Accessibility Project to Improve Voice AI for People With Disabilities
Amazon, Apple, Google, Meta, and Microsoft are partnering with the University of Illinois (UIUC) to form the Speech Accessibility Project and upgrade AI understanding of people with disabilities or unusual speech patterns. The Speech Accessibility Project will collect and analyze speech samples from people with a broad array of speech impairments to train new AI models capable of understanding them as well as anyone else.
The UIUC will helm the Speech Accessibility Project, including paying volunteers with ALS, Parkinson’s Disease, and other degenerative illnesses affecting speaking to contribute voice samples. They will all have their voices anonymized and put into a dataset for training AI models to understand people with those speech impairments. The project is starting in American English, though it could expand to other languages in the future.
“The option to communicate and operate devices with speech is crucial for anyone interacting with technology or the digital economy today,” UIUC professor and project leader Mark Hasegawa-Johnson explained. “Speech interfaces should be available to everybody, and that includes people with disabilities. This task has been difficult because it requires a lot of infrastructure, ideally the kind that can be supported by leading technology companies, so we’ve created a uniquely interdisciplinary team with expertise in linguistics, speech, AI, security, and privacy to help us meet this important challenge.”
Amazon timed some of its own accessibility news to the beginning of the Speech Accessibility Project. THe company shared a series of new videos showcasing how people with disabilities can employ its accessibility features with Alexa, Fire TV, Kindle, and other products. It also revamped the website highlighting its accessibility features.
“More than one billion people worldwide have a disability,” Amazon Devices director of accessibility Peter Korn explained. “What I love most about this work is the authentic representation—how real people with real disabilities are featured, highlighting how our products and services can help customers with disabilities live the life they want. At Amazon, we strive to build accessible products alongside people with disabilities, to benefit everyone. And with our redesigned accessibility website, we’re working to make these offerings easy to discover, too.”
The new project and related features build on previous work by tech companies to widen speech recognition to encompass those with disabilities. Google began testing Project Relate for that capability last year, founded on Google’s Project Euphonia for training assistants to understand people with unique speech patterns. Meanwhile, Alexa can be directly controlled by those with speech impairments thanks to Amazon’s partnership with speech recognition technology startup Voiceitt, which also offers an iOS app to help people with atypical and impaired speech better communicate with the AI and other people. Apple has also dipped its toe in the field, researching techniques to help Siri know when someone is stuttering to compensate and ensure it doesn’t interrupt or misunderstand the user.