New York experiments with AI to dial down the volume

Top Stories

New York experiments with AI to dial down the volume
New York is one of the loudest cities in the world.

new york - A $4.6 million project is using machine learning technology and sensors to build a sound library

  • Follow us on
  • google-news
  • whatsapp
  • telegram

Published: Mon 26 Jun 2017, 7:49 PM

Last updated: Mon 26 Jun 2017, 9:52 PM

Car horns, sirens, drilling, jet overflights and restaurants where diners have to yell to be heard - New York is one of the loudest cities in the world.
But America's most populous metropolis, known as 'The City That Never Sleeps', has launched a unique experiment seeking to provide New York with the technology to dial down the volume and address noise pollution.

The five-year, $4.6 million project - the brainchild of researchers at New York University, working in concert with city residents and city hall - is using machine learning technology and sensors to build a sound library.

The idea is to record the full panoply of noises in the city of 8.5 million residents and use artificial intelligence so that machines can recognise sounds automatically, ultimately giving authorities a way to mitigate noise levels.

"It is like living in the middle of a soccer stadium sometimes," says Gregory Orr, a filmmaker from Los Angeles who has lived in New York for 19 years. "Even the squirrels have to chirp louder in the city in order to be heard over the din," he jokes.

Juan Bello, head of the 'Sounds of NYC' project and associate professor of music technology at NYU, says noise is "consistently the number one civil complaint" to the city's 311 telephone hotline for non-emergency services, instituted in 2003.

Researchers installed the first sensor boxes, which transmit data through wifi, on New York University buildings in Greenwich Village.

They're now installing sensors across Manhattan and Brooklyn at spots selected for their diverse sounds. By the end of the year, there should be 100 in place.

"There are plenty of studies that show that noise has a tremendous impact on health, both short-term and long-term," says Bello, citing heart conditions, hearing loss and hypertension, which then have a significant economic impact.

Educational performance is also shown to suffer among children subjected to high noise levels.

In Manhattan, Bello says the effects are amplified by skyscrapers, which form "canyons of sound" and make everything louder.

"A lot of the sounds that you get in New York would not be so loud in other places, because of the specifics of the topology of the city," he tells AFP.

That was the concept from which the project was born, and it is being financed by the National Science Foundation.

The sensors are programmed to record no more than 10 consecutive seconds to avoid eavesdropping on conservations and posing confidentiality problems.

Researchers hope to index thousands of sounds which, with the help of New Yorkers, will be carefully annotated and help computers identify the source of nuisance sound immediately. It would then be over to the city to do what it can to limit it.

The problem is clear. But the solutions might still be some way off. How, for example, do you deal with something as short-lived and unpredictable as honking?

"We have to get more creative," says Bello. Today, it can take authorities five or six days to deal with a noise complaint, and requires the intervention of one of 50 specialist inspectors, says Bello. After so long, the problem has often disappeared.

Those who call the police on noisy neighbours can be given short shrift by officers who have more pressing priorities.

The first results gathered by Bello's team tend to confirm that the problem is under-reported, that there are more noise violations than the 311 log seems to suggest, at least for sounds linked to construction.

New York may not be the only loud city in the world, but Bello calls it "a perfect laboratory" to test solutions that can be adopted and transferred "to many other places in the US and around the world."

"That's ultimately the objective," he says. "We will generate a core set of technologies that can be applied to this problem anywhere." - AFP


More news from