Tackling Noise Pollution with Machine Listening
Original Source Here
Tackling Noise Pollution with Machine Listening
How audio analytics can help cities mitigate noise problem
Common sources of noise pollution
Sound is a natural part of our surroundings. But when sounds become unwanted and loud, they can turn into noise pollution. Noise pollution is considered to be any unwanted or disturbing sound that affects the health and well-being of humans and other organisms.
Sound is measured in decibels. Noises are considered to be at an acceptable level if they are between 40 and 60 decibels, or match the ambient background noise — whichever is higher. Any sound above acceptable levels is generally considered noise pollution, and sounds that reach 85 decibels or higher can harm a person’s ears.
Examples of typical noises include: city traffic (70 decibels), lawn mowers (90 decibels), subway trains (90 to 115 decibels), and car horns (110 decibels). Other major contributors to the urban noise include loud music and construction of city streets and buildings.
Noise pollution is more dangerous than we think
Noise pollution has been classified as one of the main environmental threats to public health. WHO defines noise as an underestimated threat that can cause a number of short- and long-term health problems. The most common health problem it causes is Noise Induced Hearing Loss (NIHL). Exposure to loud noise can also cause high blood pressure, heart disease, sleep disturbances, and stress. These health problems can affect all age groups, especially children.
Many children who live near noisy airports or streets have been found to suffer from stress and other problems, such as impairments in memory, attention level, and reading skill.
In the European Union alone, more than 100 million people are affected by hazardous noise levels from vehicle and aircraft traffic. In light of all the evidence on the relationship between environmental noise and specific health effects, European Commission unveiled an action plan in May 2021 to reduce pollution at source that includes a target for improving noise pollution.
How do we validate the noise source then?
Although policymakers and authorities have been aware of the seriousness, noise abatement hasn’t been easy due to the challenges with noise monitoring methods.
The typical need for measurements is to monitor the noise caused by a noise source (e.g. an airport or an industrial plant) in a residential area. However, also other noise sources exist and the captured noise level is usually a result of a combination of the target and interfering sound sources: wind-generated, cars, and birds being examples.
Then comes the validation problem: Most noise monitoring devices record only the noise levels — which makes validation impossible. How does one know if the 100-decibel record is coming from the dog barking near the monitoring device or an airplane 100 feet away? One common method to ensure the noise was caused by the original source is listening through all the samples afterwards which requires an unreasonable amount of resources.
With lack of actionable evidence of violation, liability is not clear and law enforcement naturally becomes accordingly impractical.
Making noise monitoring intelligent with ML
A considerable amount of manual work can be saved by automatically validating sound sources using Sound Identification, a machine learning task where you input some sound to a machine learning model to categorize it into predefined categories such as dog barking, car horn and so on.
At Cochl, we are developing algorithms capable of learning a sophisticated noise source classifier for an arbitrary scenario, simply using relevant annotated recordings as training material. With a network of sensors deployed around the city —being integrated into cameras, network audio or devices under street lights, Cochl’s solution continuously listens for and analyzes your target noise.
Furthermore, to alleviate the privacy issues concerning the continuous audio capturing and storage, all the computation for analysis and processing is performed on the device and only the analyzed result is transferred.
What measures can be done at the city level
Based on quantitative assessment of environment noise, policy makers can materialize plans to implement mitigation measures. Mitigation measures including re-designing city streets to moving it away from residential buildings, or installation of rail dampers and noise barriers can have direct impact on the residents’ quality of life in the long term.
Some countries like Switzerland have developed a national action plan on noise abatement. The strategic priorities include an increase in noise mitigation measures at source, the promotion of quiet and recreational areas in settlement developments and an improvement in noise monitoring and public awareness.
Noise monitoring can also hugely help with the law enforcement efforts by discovering and deterring of individuals suspected of law violation. For example, it can help identify illegal vehicle speeding that make skid noise or vroom on the city roads or people playing music with loudspeakers.
We are now looking at an increasing number of cities deploying sound monitoring for public safety to detect emergencies like gunshot, screaming and glass-break. We believe we will soon begin to witness wider application of machine learning in smart cities — from emergency detection to noise pollution reduction and law enforcement.
AI/ML
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot
via WordPress https://ramseyelbasheer.io/2021/06/01/tackling-noise-pollution-with-machine-listening/
