Royal Philips, High Tech Campus 36, Eindhoven, 5656 AE, The Netherlands.
Human-Technology Interaction, Eindhoven University of Technology, Eindhoven, 5600 MB, The Netherlands.
J Acoust Soc Am. 2018 Mar;143(3):1407. doi: 10.1121/1.5026619.
The effect of irrelevant sounds on short-term memory was investigated in two experiments using noise-vocoded speech stimuli (NVSS). Speech samples were systematically modified by a noise-vocoder and a set of stimuli varying from amplitude-modulated white noise to intelligible speech was created. Eight NVSS conditions, composed of 1-, 2-, 4-, 6-, 9-, 12-, 15-, and 18-bands, were used as the distracting stimuli in a digit-recall task next to the speech and silence conditions. The results showed that performance decreased with the number of frequency bands up to the 6-bands condition, but there was no influence of number of bands on performance beyond six bands. The results were analyzed using four acoustic metrics proposed in the literature: the frequency domain correlation coefficient (FDCC), the fluctuation strength, the speech transmission index (STI), and the normalized covariance measure (NCM). None of the metrics successfully predicted the results. However, the parameter values of the FDCC, the STI, and the NCM indicated that a prediction model for irrelevant sound effect should account for both temporal and spectral features of the irrelevant sounds.
在两项使用噪声编码语音刺激（NVSS）的实验中，研究了无关声音对短期记忆的影响。通过噪声编码器对语音样本进行了系统的修改，创建了一套从调幅白噪声到可理解语音的刺激。在一个数字回忆任务中，除了语音和静音条件外，还使用了由 1、2、4、6、9、12、15 和 18 个频带组成的 8 个 NVSS 条件作为干扰刺激。结果表明，随着频带数量的增加，性能下降到 6 个频带的条件，但在 6 个频带之外，频带数量对性能没有影响。使用文献中提出的四个声学指标（频域相关系数（FDCC）、波动强度、语音传输指数（STI）和归一化协方差度量（NCM））对结果进行了分析。这些指标都没有成功地预测结果。然而，FDCC、STI 和 NCM 的参数值表明，无关声音效果的预测模型应该同时考虑无关声音的时间和频谱特征。
