clock menu more-arrow no yes

Google’s Project Euphonia helps make speech tech more accessible to people with disabilities

New, 1 comment

Improving speech recognition software for people with voice impairments

The goal of Google’s new Project Euphonia.
Credit: Google

Voice interfaces are more common than ever, but they’re not equally accessible. For example, if you have a speech disorder cause by a neurological impairment like ALS or multiple sclerosis, then using Google Assistant or Amazon’s Alexa will be off-limits to you. In today’s tech environment, that means missing out on a lot.

That’s why Google is launching a new initiative to make speech technology more accessible to those with disabilities. It’s called Project Euphonia, and it incorporates a wide array of research directions, alongside collaborations with nonprofits and volunteers.

The biggest focus of Euphonia will be collecting more voice data from people with impaired speech. This is intended to remedy the problem of AI bias created by limited training data. Because speech software like Google Assistant is built to respond to the majority of voices, it doesn’t work for those in the minority — such as people with voice impairments.

To fix this, Google is asking people around the world to submit their voice sample. It hopes by collecting this data it can improve its algorithms, eventually integrating the updates into Google Assistant. (If you’re interesting in submitting your voice or the voice of someone you know, the contact form can be found here.)

In addition, the company is working on new interactive AI systems that recognize actions like gestures and facial expressions. That would mean people with severe disabilities who cannot speak at all could also use technology like smart home speakers and lights.

At I/O, the company also unveiled a prototype app called Live Relay, which uses on-device speech recognition and text-to-speech conversion to help anyone who can’t hear or speak hold a phone conversation.

It would be helpful to individuals with hearing impairments, letting them see instant transcriptions of what their conversational partner is saying, but could also help users who just don’t want to speak. If you’re in a public place for example, you can just type into your phone and what you write will be turned into speech at the other end.

However, Live Relay is only a prototype for now which Google says is “still in the research phase.” There’s no firm timetable for when it might be released to the public.

“Fundamental AI research which enables new products for people with disabilities is an important way we drive our mission forward,” said Google CEO Sundar Pichai onstage. “[These projects] will ultimately result in products that work better for all of us. It’s the perfect example of what we mean by building a more helpful Google for everyone.”