NYU Calling on Citizen Scientists to Train Machine models on Noise Pollution Sounds

By: Marie Donlon

Researchers at New York University (NYU) are calling on citizen scientists to participate in a study addressing urban noise pollution.

The Sounds of New York City (SONYC) project, a collaboration between New York University (NYU), the City of New York and Ohio State University, is launching a citizen scientist initiative that would enable artificial intelligence (AI) tech to identify the sounds that are lending to unhealthy noise pollution levels throughout the city.

Currently in its third year of a five year study, researchers have already collected two years’ worth of sound data from acoustic sensors placed in high-noise locations throughout the city. At the citizen scientist phase of the study, volunteers are asked to identify and label 10-second sound clips like the sounds of a jackhammer or an ice cream truck as captured by those acoustic sensors and held in the Zooniverse citizen science web portal. There, participants train the machine listening models to recognize the sounds autonomously while also helping researchers to categorize the sound data. Such information, according to researchers, may offer city officials even more precise data about noise pollution causes and possible solutions.

“It’s impossible for us to sift through this data on our own but we’ve learned through extensive research how to seamlessly integrate citizen scientists into our systems and subsequently, advance our understanding of how humans can effectively train machine learning models,” said lead investigator Juan Pablo Bello, a professor of computer science and engineering at the NYU Tandon School of Engineering, director of NYU’s Center for Urban Science and Progress (CUSP) and director of the Music and Audio Research Lab (MARL) at the NYU Steinhardt School of Culture, Education, and Human Development.

Currently, the initiative has attracted 600 participants to help train the machine listening models. However, researchers estimate that they will need roughly 3,500 volunteers in all. To join the effort, go to the Zooniverse citizen science web portal.

This article originally appeared on Engineering360.