The Speech Accessibility Project is an interdisciplinary initiative with one shared goal: to improve speech technology for people with a range of diverse speech patterns and disabilities. Led by the University of Illinois Urbana-Champaign with support from Amazon, Apple, Google, Meta, and Microsoft, the project brings together technologists, academic researchers, and community organizations to create the diverse speech data needed to make spoken interfaces more accessible for everyone.
Speech recognition is a powerful tool to make technology more accessible in our daily lives, especially for people with disabilities. However, it is not fully accessible for many people with speech disabilities.
Speech recognition is powered by machine learning, and without diverse, representative data, ML models cannot learn how to understand diversity of speech. This project aims to change that by creating the dataset needed to More effectively train these machine learning models.
Instead of separate and duplicative initiatives by different companies and research teams, the groups will collaborate on this project to gather a set of high-quality, representative speech samples that will help accelerate the technologies that support these communities of people with diverse speech patterns.
Millions of people will benefit from diverse and inclusive speech recognition. This may include but isn’t limited to people with Amyotrophic Lateral Sclerosis, Parkinson’s disease, cerebral palsy, Down syndrome and a wide range of medical and non-medical conditions that affect speech.
The companies supporting the Speech Accessibility Project – Amazon, Apple, Google, Meta and Microsoft – each have longstanding accessibility commitments that include delivering products, services, and experiences for people from a wide range of communities and backgrounds.
This one-of-a-kind collaboration is rooted in the belief that inclusive speech recognition should be a universal experience. Working together on the Speech Accessibility Project is a way to provide the best and most expedient path to inclusive speech recognition solutions.
UIUC is leading the Speech Accessibility Project, managing data collection and curation, privacy, authenticity, and equitable access for researchers. With programs in computer science, speech, hearing, linguistics, and accessibility, UIUC not only has a historic commitment to disability inclusion, but the expertise needed to make the project a success. The project is happening at UIUC’s Beckman Institute for Advanced Science and Technology, which brings together researchers across disciplines to solve problems and break technological barriers.
University and industry partners are committed to working closely with disability community organizations to identify areas of need, recruit participants, review the participation and consent process, and get feedback on how the project can be improved over time.
The study is currently screening potential participants via the Speech Accessibility App. They are particularly recruiting U.S. residents with Parkinson’s in the first phase of the project. Each participant will work with a mentor from LSVT Global to record speech samples.
Volunteer participants record speech samples, for example, by reading text prompts, or by answering questions out loud. The questions include things like, “How would you spend a rainy day?” or “What are your hobbies?”
The texts that volunteer project participants read, and the questions they respond to, were developed in collaboration with focus groups and disability community organizations in order to make sure the study is collecting data that will help researchers create datasets to more effectively train speech recognition technologies.
The Speech Accessibility Project is a multi-year research initiative. However, each of the companies supporting the Speech Accessibility Project is committed to leveraging project data to make improvements within their respective voice recognition products and services.
UIUC maintains a secure, private and encrypted tool to store and access the Speech Accessibility Project’s research data. Participant data will be de-identified to protect each participant’s privacy and personal information.
The goal for this project is to support a range of efforts to make voice recognition technology more useful for people with diversity of speech patterns or disabilities. The de-identified data will be made available to researchers and developers.
At this link, you will sign-up to create an account. Then, you’ll have two options for getting started:
A mentor from LSVT Global will contact you, evaluate your speech and see if you qualify for the project. If you do, they’ll guide you through the informed consent process to get you started.
Or, you can record a few sentences to help the team evaluate your speech patterns and whether they fit the profile the study is currently seeking. After they have had a chance to listen to your speech, you will be emailed and prompted to complete informed consent documents before continuing with the study.
If you need help, you can find instruction under the help tab at the link, or email your questions to speechaccessibility@beckman.illinois.edu
Comments