목적: This study quantified the impact of ambient noise on kiosk-based automated audiometry and evaluated strategies for stimulus intensity adjustment during hearing assessments. 방법:Experiments were conducted in the Department of Otorhinolaryngology at Wonju Severance Christian Hospital. Three experiments were performed to evaluate the influence of ambient noise. Experiment I measured environmental noise levels using a sound level meter (low, mid, high). Experiment II used an artificial ear with free-field and reference microphones to simulate human listening under three test setups involving headset and open-field configurations. Experiment III compared kiosk-based automated audiometry with conventional audiometric testing to determine threshold deviations. 결과:Equivalent continuous noise levels (LZeq) ranged from 66.6–75.6 dB SPL, showing the largest fluctuations during outpatient hours in Experiment I. Artificial ear results indicated higher noise with the headset applied, and a gradual spectral decline from low to high frequencies in Experiment II. In Experiment III, pure-tone thresholds in kiosk-based automated audiometry conditions were elevated around 0.5–1 kHz compared to traditional audiometry. However, under the 40 dB HL screening condition, high sensitivity (88.14–94.37 %) and specificity (77.42–94.74 %) were observed in the 2–8 kHz frequency range. Furthermore, the measurement results showed a high correlation across all five frequencies (R ≥ 0.45; p < 0.05). 결론:In summary, this study first quantified environmental noise levels (Experiment I) and evaluated passive noise reduction (Experiment II), before comparing the automated kiosk system with traditional audiometry (Experiment III). Kiosk-based automated audiometry showed a strong correlation with traditional audiometry across all frequency ranges, even in the presence of background noise. Although the relatively high low-frequency noise contributed to elevated pure- tone thresholds at 0.5–1 kHz, the observed strong correlation suggests that further refinement, such as implementing effective masking or correlation-based noise compensation algorithms, is highly promising for achieving reliability in future studies.