South Dakota researcher receives NSF grant for speech recognition study
A $200,000 award will allow an assistant professor to test how different hardware impacts the accuracy of AI transcription tools.
By Samantha Reed ·
Nirmalya Thakur, an assistant professor at South Dakota School of Mines and Technology, has been awarded 200,000 ACCESS Credits to fund a new study on artificial intelligence speech recognition.
The funding comes from the National Science Foundation and will provide Thakur’s team with access to advanced computing resources. These resources are often unavailable at smaller universities.
The research team will investigate how different graphics processing units affect the performance of automatic speech recognition software. The goal is to determine if specific hardware influences how accurately and consistently AI systems transcribe spoken words.
Speech-to-text tools are widely used to generate captions and transcripts. Improving their reliability could benefit people who are deaf or hard of hearing, as well as others who depend on these technologies for daily communication.
Thakur stated that the award will support his work in artificial intelligence and high-performance computing. He aims to identify factors that influence transcription quality to make the technology more dependable.
The project was selected through the Explore category of the ACCESS program. Launched in 2022, the program, which stands for Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support, provides computing resources to researchers across various scientific fields.
