Moving into the next phase for voice-activated technology accessibility
KALAMAZOO, Mich.—An interdisciplinary group of researchers from Western Michigan University and Michigan State University has received an National Science Foundation (NSF) Convergence Accelerator grant to make voice-activated artificial intelligence (AI) more accessible to those who stutter or have other speech differences.
An additional $5 million award will continue the second phase of their work over the next three years, says WMU's Dr. Hope Gerlach-Houck, assistant professor of speech, language and hearing sciences and co-principal investigator for the project, which received its first NSF grant in 2022.
"Speech recognition technology is becoming more and more important in everyday life," she says. "This technology is embedded in our healthcare system, in the cars we drive and in the phones we use. The consequences of not being able to access voice AI can range from being annoying to life-threatening. What if, for example, you're in a car accident and can't reach your phone?"
With voice AI technology expected to quintuple by 2030, there is an urgency about the team’s work. An estimated 1% of the population stutters—which amounts to more than 3 million people in the U.S. and 80 million people worldwide. Much of Gerlach-Houck's work has focused on reducing disparities in health and quality of life that are often associated with stuttering.
As co-PI on the project, Gerlach-Houck has contributed by designing studies to learn about the stuttering community’s experiences with voice AI.
"We surveyed adults who stutter about their experiences with voice AI and asked them to describe and rate barriers and facilitators to voice AI use," she says. "Seventy-five percent reported that they experienced difficulty using voice AI at least some of the time, with 38% reporting frequently or always experiencing difficulty."
Participants estimated that voice AI systems misinterpreted their speech 60% of the time. But they still reported a desire to use voice AI technology more, with 79% indicating they would use voice AI more if it worked more accurately.
There's an app for that
The app is still under development, but the team has used it to collect speech samples and get input from people who stutter. “Our team has connections with so many people who stutter,” said Mahapatra. “As we develop the app, we will constantly engage them to make sure the technology addresses their needs.”
Once fully developed, HeardAI will include flexible solutions for individuals interacting with voice AI systems like Siri, developers creating accessible voice AI services, and organizations relying on accurate ASR transcripts.
During the second phase of the project, the group has two priorities beyond delivering the HeardAI application itself. They also hope to create a “test bed” that other developers can use to help make their own applications work with stuttered speech. And finally, the team will create accessibility standards to guide the development of voice AI technology going forward.
In addition to her research role, Gerlach-Houck is also the publicity and marketing lead on the project. She was heavily involved in creating the project video and is currently working on launching a website for the project.
She is optimistic that the outcomes of this project will positively impact individuals with speech differences, those that she has focused on throughout her career. "People who stutter deserve to live their lives as fully as everyone else," she says. "And making voice AI accessible is a critical piece of the puzzle."
More information
- From WMU News: Making AI more accessible for people with speech differences
- From the MSU Engaged Scholar eNewsletter: Building Voice AI That Works for Everyone
For more WMU news, arts and events, visit WMU News online.