A sound minimization system for enhanced indoor acoustics

Abstract

Indoor noise pollution has become a critical public health concern particularly due to rapid urbanization. However, existing Passive Noise Control (PNC) solutions inherently compromise Indoor Environmental Quality (IEQ) by impeding air flow and blocking sunlight, while the traditional Active Noise Control (ANC) methods remain cost-prohibitive for widespread deployment. This project introduces an indoor sound minimization system that utilizes a Machine Learning-based ANC framework. The system integrates microphones, speakers and microcontrollers with a Convolutional Recurrent Network (CRN) algorithm to predict and generate effective real-time anti-noise signals. Performance evaluation of a working prototype in diverse acoustic environments demonstrates an average noise reduction of 9.4 decibel (dB) in the frequency range of 20 to 2000 Hertz (Hz), achieved within 80 milliseconds (ms). The system prioritizes scalability and low power consumption, positioning it as a potentially viable and sustainable acoustic solution for residential homes and healthcare facilities.