Matched standard samples method in laboratory listening tests for annoyance perception

Subjective noise assessment, expressing people's feelings about noise qualitatively or quantitatively, plays a crucial role in noise control and management. Among established methods for assessing sound perception characteristics, the anchor rating scale stands out for its low requirements for participants, moderate workload, and high accuracy. However, inconsistencies in anchor selection can lead to incommensurability between subjective assessment results obtained from various sample sets. While the standard sample method (SSM) addresses this issue, its assessment accuracy relies on the noise characteristics being assessed. This paper seeks to remedy this limitation by conducting comparative listening tests for three versions of SSM that compare their performance to the anchor rating scale and proposing a reliable engineering application guideline. The comparative study demonstrated that white noise outperformed pink noise in assessing noise annoyance. Furthermore, both pure tone and white noise presented different advantages in assessing the noises with prominent perceptual difference, which is related to the inherent characteristics of the noise samples themselves. To ensure the selection of the appropriate standard sample type for obtaining accurate subjective results when employing the standard sample as the anchor to assess annoyance caused by a group of noise, the matched standard samples method (MSSM) was proposed. A decision function was developed based on psychoacoustic parameters like sharpness, roughness, tonality, etc. The MSSM automated the selection of the most appropriate standard sample type for a tested noise set, utilizing a decision function as the judgment basis. Within the chosen standard sample type, the one with a sound level closest to the median of the tested noises was the optimum standard sample for annoyance assessment of the noise set. This “matched” identified an optimum standard sample for a specific noise set, ensuring close alignment with the tested noise's annoyance-generating mechanism and median sound level. The listening test confirmed that the MSSM exhibited a superior correlation with the anchor rating scale method and higher accuracy compared to the SSM. The MSSM solidifies the scientific basis of standard sample selection and offers a reliable guideline with as few uniform scales as possible for assessing noise auditory perception in engineering applications.