Google has made tremendous strides in speech recognition technology. However, this tech currently doesn’t fully cover individuals with speech impairments. The company wants to rectify this with a new initiative called “Project Relate.”
The company is now accepting volunteers for Project Relate to contribute towards multi-year research with the goal of improving Google’s speech recognition algorithms. Google ultimately wants to make sure Assistant and the rest of its voice-based features are fully capable and inclusive of people with neurological conditions or speech impairments.
“Project Relate is a continuation of years of research from both Google’s Speech and Research teams, made possible by over a million speech samples recorded by participants of our research effort,” Google said in a blog post (via).
Google is also continuing its work on Project Euphonia which started in 2019
The sign-up page is now live at g.co/ProjectRelate. Google says volunteers need to be 18 or older, and “have difficulty being understood by others” to be eligible.
Project Relate is currently available for English speakers in the U.S., Canada, Australia, and New Zealand. Volunteers will need an Android smartphone (Android 8.0 or newer) and a Google account to be eligible.
“I’m used to the look on people’s faces when they can’t understand what I’ve said,” Google Brand Manager, Aubrie Lee said. “Project Relate can make the difference between a look of confusion and a friendly laugh of recognition.”
When confirmed, volunteers have to record up to 500 phrases to train the system. This process can take anywhere between 30 to 90 minutes. Following the submission of voice samples, the Relate app opens up three core features – Listen, Repeat, and Assistant.
Listen basically transcribes everything the user says in real-time. Repeat involves restating what the user said in a synthesized and clear voice, ideal for a noisy or public environment. These two features help people with speech impairments hold regular conversations with those around them. Lastly, Assistant effectively works as it usually does, letting you open the camera, make calls, and so on with a voice command.
Google is training its algorithms to recognize gestures and sounds to assist people who cannot speak. Separately, the company is also continuing work on Project Euphonia. This came into existence in 2019 and involves collecting voices from people with speech impairments to improve its AI algorithms.