Framework for Acoustic Comfort Analysis in Digital Twins

Abstract

Acoustic comfort, a critical yet often overlooked aspect of indoor environmental quality, plays a significant role in occupant health and productivity. Unlike other comfort dimensions, such as thermal or lighting, measuring acoustic comfort remains challenging due to its subjective nature and the complex interplay of physiological and psychological factors. Current approaches to assessing acoustic comfort in indoor environments often overlook the content of sound, despite its potential to be a decisive factor. For instance, the perceived comfort of listening to music at high sound levels differs significantly from that of hearing construction noise, even at lower sound levels. This research proposes a novel framework for integrating acoustic comfort analysis in the digital twin environment, which comprises psycho-acoustic metrics, sound event classification, and predictive analytics. The implemented system leverages sensor data, a sound event classification neural network, and advanced visualization methods to enable real-time and historical acoustic analysis. Privacy concerns are addressed through a privacy-by-design approach, ensuring data security by processing audio on the edge devices without storing raw sound. A case study in an office environment demonstrates the framework's effectiveness in monitoring and improving acoustic conditions. Microphones connected to edge devices classify sound events and calculate soundwave parameters such as relative sound pressure levels while integrating results into the digital twin.