Google is developing an app for Android to facilitate communication for people with speech difficulties. Project Relate is the name of the project, which aims to generate an automatic transcript of a spoken speech while synthesizing the information so that it is more easily understood by the speaker.
Project Relate is a Google initiative to provide communication tools for people with speech impairment
The seed of this idea comes from Project Euphonia, a Google initiative born in 2019 that aims to provide communicative tools to people with some kind of speech impediment. This project, based on an Artificial Intelligence solutionwas led by Dimitri Kanevsky, a Google researcher who has impaired speech due to an illness. Aubriee Lee, who works in the marketing department, is another of the beneficiaries of this technology. She herself appears in the promotional video explaining the Google project.
Speech recognition systems require a large number of recordings in order to learn to correctly interpret what is being said. In general, these databases include mainly recordings of people with correct speech. For this reason, it is difficult to make recognition systems compatible with impaired speech or heavily accented speech.
To overcome this obstacle, Google has been collecting recordings of the speech of people with difficulties. In this way, the company has generated a database of about one million speech samples from volunteers.
Based on this information, Google has been able to train its aI-guided speech recognition system. To continue development, Google is recruiting volunteers to use the app in order to address its shortcomings.
This past summer the company introduced another Artificial Intelligence tool. It is Multitask Unified Model, an algorithm that counts capable of providing answers so complex that the previous algorithm would have needed up to eight different searches.