Noise Pollution Classification Using Deep Learning

With the rise and rapid growth in industrialization as well as urbanization, noise pollution has become a significant yet often overlooked threat to our environment. Transportation, human chatter, public transportation, and air conditioners produce high level of unwanted sound. Since noise pollution...

Full description

Saved in:
Bibliographic Details
Published in:2025 International Conference on Sustainability, Innovation & Technology (ICSIT) pp. 1 - 6
Main Authors: Sharma, Aditya, Sen, Madhusmita, Marak, Agimik Sengat B., Jaiswal, Dhiraj Prasad, Shah, Rahul
Format: Conference Proceeding
Language:English
Published: IEEE 22.08.2025
Subjects:
Accuracy
Atmospheric modeling
Convolutional Neural Network
Convolutional neural networks
Deep learning
Environmentally friendly manufacturing techniques
Feature extraction
Machine Learning
MelSpectrograms
Noise
Pollution
Real-time c
Real-time systems
Working environment noise
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:With the rise and rapid growth in industrialization as well as urbanization, noise pollution has become a significant yet often overlooked threat to our environment. Transportation, human chatter, public transportation, and air conditioners produce high level of unwanted sound. Since noise pollution has no visible indicators, it gets difficult to detect, quantify, and manage effectively, which makes it a more pressing issue than other forms of pollution. Prolonged exposure to such noise levels may lead to hazardous health issues such as hearing loss, stress, cardiovascular diseases and also sleep disorders. The traditional noise monitoring techniques are often inefficient, non-scale, and, most importantly, not accessible to the general public. This research explores the Deep Learning (DL) - based approach - Convolutional Neural Networks (CNNs) for environmental noise classification. It involves converting raw audio into Mel-spectrograms using standard Python libraries enabling the developed CNN to extract and learn complex audio features. In this paper, authors have trained CNN Model to learn various noise patterns and classify them into categories which includes traffic, human chatter, animal sounds, air conditioners and many more. The authors also developed a web platform which facilities real-time noise classification and visualization, aiming to raise public awareness and noise monitoring at the community level. The proposed CNN model achieved 93.53% accuracy, demonstrating its effectiveness in classification of environmental noise types using Mel-spectrograms.
DOI:10.1109/ICSIT65336.2025.11295332