(AURN News) — Howard University and Google are working together on a new project that could help voice technology better recognize how many Black Americans speak.
The effort, known as Project Elevate Black Voices, focuses on the way speech recognition systems often fall short when it comes to African-American English.
Howard says the team has collected 600 hours of voice data from Black speakers across 32 states. That data is being used to build a high-quality speech dataset of African American English — or AAE — with the goal of improving automated speech recognition, the technology that powers voice assistants, Google Maps and YouTube’s automated transcription.
“The team has completed data collection efforts where they have created a high-quality speech dataset of African-American English (AAE) across the United States to reduce racial disparities in automated speech recognition systems,” the project’s website states.
Howard University will retain ownership of the dataset and oversee how it is used to ensure it benefits Black communities. According to the project’s website, Google can also use the data to improve its own tools and services.
The initial release of the dataset will be limited to researchers and institutions within the network of Historically Black Colleges and Universities. “This is to ensure that the data is employed in ways that reflect the interests and needs of marginalized communities, specifically African American communities, whose linguistic practices have often been excluded or misrepresented in computational systems,” the website explains.
Howard says broader access may be considered in the future, with priority given to organizations whose work supports “inclusivity, empowerment and community-driven research.”
Click play to listen to the AURN News report from Jamie Jackson:
The post Can This Howard and Google Project Make AI Less Biased? appeared first on American Urban Radio Networks.
