Prototype eases web navigation for those with visual impairments Aug. 20, 2019
by Brent Davis
Waterloo Region Record
A University of Waterloo student has helped develop a new tool to make online navigation easier for people with visual impairments.
The prototype, dubbed VERSE Voice Exploration, Retrieval, and Search merges voice-based virtual assistants on devices like phones and smart speakers with screen readers, which can read out the contents of a web page to the user.
Alexandra Vtyurina, a PhD candidate at UW’s David R. Cheriton School of Computer Science, collaborated with researchers at Microsoft and University of Washington assistant professor Leah Findlater during an internship at Microsoft Research in Redmond, Wash. last summer.
Their findings are expected to be presented at a conference on computers and accessibility in Pittsburgh in October.
“I don’t think that we talk enough about people with visual impairments, people who are blind, and we certainly should be,” said Vtyurina, who co-authored the study. “It was definitely very satisfying to be working on something that made a difference.”
Researchers surveyed 53 visually impaired people and examined which devices they use to search for online content. Many people use both virtual assistants and screen readers, but both have their limitations, Vtyurina explained.
Screen readers provide in-depth engagement with web content but can be difficult to learn to use, and can run into problems if web content is not designed to accessibility standards, she said. The various screen readers available also have different strengths, prompting some people to use multiple readers for different purposes.
Voice assistants are convenient and easy to use, but are often limited in their answers or the extent to which they can provide information beyond the first few lines of an article.
“That’s the trade-off we have between voice assistants and screen readers,” Vtyurina said.
The VERSE prototype aims to marry the best attributes from both types of devices, adding screen reader-like capabilities to virtual assistants and allowing users to search for information beyond that initial answer. The voice-driven tool can also be paired with an app for use on a phone or smartwatch.