Clustering the residential noise in apartment house based on spectral and temporal analysis

This study aimed to investigate how the residential noises cluster based on spectral and temporal characteristics of the noises in apartment houses. The type of noise was floor impact, air-borne, plumbing, transmitted outdoor unit of air-conditioner, traffic (road, rail, and air-craft noise), and construction noise. Duration of sound was edited to 5 s and sound pressure level was adjusted to be equal to be 50 dBA. K-means clustering analysis was performed to classify various sound sources, and the elbow method was used to confirm the number of clusters. As a result, residential noise sources were classified into twelves clusters according to spectral and temporal characteristics and difference in psychoacoustic parameters between clusters was also found. In addition, perceptual aspect of each cluster was investigated through sematic differential test.