THE University of Illinois (UIUC) wants to expand the accessibility of voice recognition solutions with the Speech Accessibility Project. The institution joined the apple, Google and other companies to optimize the artificial intelligence (AI) behind the technology. Thus, these systems will be more accurate in identifying the speech of people with Amyotrophic Multiple Sclerosis, Parkinson’s, Down Syndrome and the like.
The initiative comes from the University of the United States in partnership with non-profit organizations and technology companies. Among them are Amazon, Apple, Google, Meta and Microsoft. In common, all actors will help to optimize the artificial intelligence systems used for voice recognition in virtual assistants, dictations and the like.
The action aims to expand the use of these solutions. Not for nothing, the project website points out that these systems do not always recognize people with a diversity of speech patterns often associated with disabilities. As a result, many individuals in these and other communities are unable to benefit from speech recognition tools.
And that’s where the study to put an end to this bottleneck comes in.
Project will collect voice samples to train AI
The project will collect samples from individuals representing a diversity of speech patterns. For this, the researchers will recruit paid volunteers to develop a private and anonymous database. This information will then be used to train machine learning (ML) models to optimize AI solutions.
“Rather than separate and duplicate initiatives by different companies and research teams, the groups will collaborate on this project to bring together a set of high-quality, representative speech samples that will help accelerate technologies that support these communities of people with diverse speech patterns. ”, says the project website.
The University of Illinois will also be responsible for the research data. According to the institution, the information will be stored in a secure and encrypted place. “Participant data will not be identified to protect the privacy and personal information of each participant,” they explained.
Apple, Google and more will be able to use study data
The action will focus, first, on English speakers in the United States. In a question-and-answer page, the university stated that the project is a multi-year research initiative. But this does not prevent the information from being used by the companies participating in the project to optimize their systems when necessary.
“Each of the companies supporting the Speech Accessibility Project is committed to leveraging the project’s data to make improvements to their respective speech recognition products and services,” they explained.
The data will also be available to other researchers. “Unidentified data will be made available to researchers and developers in accordance with the UIUC data usage agreement,” they stated.
Currently, the project is in the pilot phase.