People use speech recognition to operate technology using their voices, in applications such as voice assistants or translation tools. Speech recognition is powered by machine learning; without diverse, representative data, machine learning models cannot learn how to understand a diversity of speech. This project aims to change that by creating a dataset representing a diversity of speech patterns, for the purpose of more effectively training these machine learning models.

The Speech Accessibility Project will collect speech samples from individuals representing a diversity of speech patterns. UIUC researchers will recruit paid volunteers to contribute recorded voice samples and will create a private, de-identified dataset which can be used to train machine learning models to better understand a variety of speech patterns. Data will be distributed to consortium partners first, then, after a short grace period and with the protection of a data use agreement, to any other organization with the interest and the ability to advance speech accessibility.

All software created by the Speech Accessibility Project will be released open-source, in the repositories of this github organization.

For more information, please see https://speechaccessibilityproject.beckman.illinois.edu/.