Researchers at the University of Pittsburgh have cracked the code of sound recognition with the help of a machine learning model. This model mimics how the brains of social animals can distinguish communication sounds like mating calls, food, and danger and respond to them accordingly. This insight provides an in-depth look into the neuronal processing systems and could be used to improve hearing aid technology or treat disorders affecting speech recognition.
Lead author and Pitt postdoctoral fellow Satyabrata Parida and senior author Srivatsun Sadagopan, Pittsburgh’s assistant professor of neurobiology, published the paper in Communications Biology.
The team used a sophisticated machine learning model to recognize the different sounds made by social animals. They then compared the brain activity of guinea pigs listening to their kin’s calls to the model. Neurons in regions responsible for sound processing were activated when hearing noises related to specific sound categories.
Guinea pigs were then trained to select between different sound categories and receive a fruit reward. To test the model to the real life behavior of animals, the team applied sound altering software, such as changing the pitch, speed, and adding noise and reverb. The results showed that the guinea pigs still maintained consistent performance despite the altered noises.
Machine learning models can be applied to humans as well as animals. To improve speech recognition or understand neurological speech conditions, the model’s accuracy must be transferred across species.
Learning the way our brains interpret sound is a valuable insight. If we can recognize how our brains detect, process, and interpret sound, we can apply this information to hearing aids to improve them and develop ways to treat disorders that affect our speech comprehension.
The company mentioned in the article is the University of Pittsburgh, one of the leading research universities in the United States. UPitt is located in Pittsburgh, Pennsylvania, with a mission to educate exceptional people, generate path-breaking scholarship, and make lasting contributions to the community, the country, and the world.
The person mentioned in the article is Srivatsun Sadagopan, Assistant Professor of Neurobiology at UPitt. He’s most well known for his groundbreaking research on sound recognition and his work to shed light on the complexities of neuronal processing.
