Audio injection based combined noises annoyance suppression: Quantification and modeling

Conventional noise control strategies reduce acoustic energy, while the Audio Injection Method (AIM) mitigates the combined noise annoyance from an auditory perspective by introducing additional sound to be added into the target noises. Previous AIM evaluations lacked direct comparability with traditional methods and often ignored the significant influence of individual participant differences (non-acoustic factors). This study investigates the quantitative evaluation and modeling of annoyance resulting from the application of AIM to substation and range hood noise. Different controllable sounds were injected, and particular emphasis was placed on incorporating non-acoustic factors of participants—specifically, their preference for the controllable sounds and their acceptance of the AIM—into the evaluation of annoyance variations. The Auditory Equivalent Noise Reduction Conversion (AENRC) method was employed, utilizing 50 dB(A) white noise as a standard sample, which converts the change in annoyance into an equivalent change in the SPL(A) of the standard sample, enabling robust quantitative evaluation and analysis of AIM annoyance suppression effectiveness. A multiple linear regression model was developed using the differences in acoustic parameters between the target noises and controllable sounds as independent variables, which significantly improved the fitting effect of the model. Additionally, cluster analysis based on the two non-acoustic factors classified participants into four distinct groups. Subsequently, a mixed-effects modeling approach revealed the significant impact of these non-acoustic factors on the annoyance of the combined sound. Key findings include: (1) The AENRC facilitates reliable quantitative evaluation of AIM's annoyance suppression effect, potentially reducing experimental variability. (2) Music was the most effective controllable sound, yielding an annoyance suppression effect equivalent to a 6.35 dB(A) reduction in the reference 50 dB(A) white noise. (3) Non-acoustic factors significantly influenced the combined noise annoyance. A statistically significant difference was observed between the participant clusters exhibiting the lowest and highest acceptance/preference (Cluster 1 vs. Cluster 4), with a mean annoyance difference of 1.76 scale units and a corresponding 35.2 % difference in the annoyance suppression effect.