Estimating the subjective severity of laptop fan abnormal sounds using psychoacoustic parameters

Abstract

The demand for acoustic comfort in domestic appliances has seen a remarkable surge, with the laptop, a prevalent device in modern homes, often being a source of undesirable noise. Such noise, especially abnormal fan sounds, can adversely affect user comfort during operation. Manufacturers are thus keen to differentiate these abnormal sounds from the routine noise of laptop fans to ascertain potential issues. This study explored the relationship between the severity of abnormal laptop fan sounds and key psychoacoustic metrics through two subjective experiments. In Experiment 1, a group 24 participants evaluated 72 laptop fan noise stimuli, which were generated from records of laptop fans operating at 12 distinct speeds ranging from 2700 to 4900 rpm in 200 rpm increments. In Experiment 2, another group of 24 participants estimated 64 stimuli characterized by pronounced abnormal sounds across various fan speeds, with these sounds adjusted to have equal loudness at each rpm. The contribution of psychoacoustic metrics, i.e., loudness, sharpness, roughness, fluctuation strength, and tonality (hearing model), on the perceived severity of the abnormal sounds, was analyzed by partial correlation analysis. Psychoacoustic analysis revealed loudness dominance in abnormal noise perception, with the derived multivariate linear regression model demonstrating determination coefficient (R² = 0.914). Cross-experimental validation established complementary mediation through roughness and sharpness modulations, suggesting their systematic incorporation with loudness as triaxial diagnostic parameters for enhanced anomaly detection in engineering applications.