Clinical & Experimental Audiology, Amsterdam University Medical Centres location AMC, Amsterdam, the Netherlands.
Pento Audiological Centre, Utrecht, the Netherlands.
Trends Hear. 2023 Jan-Dec;27:23312165231192304. doi: 10.1177/23312165231192304.
There are large interindividual variations in preference for strength of noise-reduction (NR). It is hypothesized that each individual's tolerance for background noise on one hand and signal distortions on the other hand form this preference. We aim to visualize and analyze this so-called trade-off between noise attenuation and signal quality. Dutch sentences in stationary background noise were processed with different NR strengths. We used an NR algorithm that allows us to separate the positive effects of NR (noise attenuation) from the negative effects (signal distortion). Stimuli consisted of speech in noise with different degrees of (1) background noise, (2) signal distortions, or (3) both (i.e., realistic NR at different NR strengths). With pairwise comparisons, participants chose which stimulus they preferred for prolonged listening. Twelve listeners with mild to moderate hearing loss participated in the study. For all participants, a trade-off between noise attenuation and signal quality was found and visualized. The strength of preference was estimated with the Bradley-Terry-Luce choice model and was different for all individuals but in the same order of magnitude for distortion effects and noise effects. Strength of preference of realistic NR was smaller by a factor of ten. This study used a unique setup to capture the individual trade-off between noise attenuation and signal quality in NR. Disturbance from signal distortions is as important as disturbance from background noise for determining preference for NR strength. Individual listeners differ in their sensitivity to both factors and as a consequence in their preferred NR strength.
个体之间对降噪（NR）强度的偏好存在很大差异。据推测，个体对背景噪声的容忍度和对信号失真的容忍度共同构成了这种偏好。我们旨在直观地分析和研究这种所谓的噪声衰减和信号质量之间的权衡。在静止背景噪声中处理荷兰语句子时，使用了不同的 NR 强度。我们使用了一种 NR 算法，可以将 NR 的积极影响（噪声衰减）与负面影响（信号失真）分开。刺激物由噪声中的语音组成，具有不同程度的（1）背景噪声、（2）信号失真或（3）两者（即不同 NR 强度下的真实 NR）。通过两两比较，参与者选择他们更喜欢长时间听的刺激。12 名听力轻度至中度受损的听众参与了研究。对于所有参与者，都发现并直观地呈现了噪声衰减和信号质量之间的权衡。偏好强度用 Bradley-Terry-Luce 选择模型进行估计，对所有个体来说都不同，但失真效应和噪声效应的顺序相同。真实 NR 的偏好强度小了一个数量级。本研究使用了一种独特的设置来捕捉 NR 中噪声衰减和信号质量之间的个体权衡。信号失真引起的干扰与背景噪声引起的干扰对确定对 NR 强度的偏好同样重要。个体听众在对这两个因素的敏感性以及因此对其偏好的 NR 强度方面存在差异。
