Glasberg B R, Moore B C
Department of Experimental Psychology, University of Cambridge, England.
J Acoust Soc Am. 1994 Jul;96(1):134-44. doi: 10.1121/1.410473.
Growth-of-masking functions were obtained for sinusoidal signals at three frequencies (fs), 0.25, 1.0, and 4.0 kHz, using maskers that were always higher in frequency than the signal. Five different maskers were used, chosen so as to evaluate the influence of temporal fluctuations in the maskers and of combination products produced by the interaction of components within the maskers: A sinusoid (S); a narrow-band noise with a bandwidth of 16 Hz (N); a noise with a slightly wider bandwidth equal to 0.75 times the equivalent rectangular bandwidth (ERB) of the auditory filter at each fs (W); a noise with a very wide bandwidth equal to 0.4fs and with lower cut-off frequency and spectrum level matched to those of masker W (V); and a sinusoidal carrier frequency modulated by a noise, and matched in bandwidth and center frequency to masker N (F). The center frequencies of maskers S, N, W, and F were either 1.1fs or 1.2fs. Masker S generally produced the smallest amount of masking and gave growth-of-masking functions with the shallowest slopes (much less than unity). Results were similar for maskers N and F; both produced slightly more masking than masker S and growth-of-masking functions with slightly greater slopes than masker S. Maskers W and V produced more masking than the other maskers, and gave growth-of-masking functions with steeper slopes. For all maskers, the slopes of the growth-of-masking functions were lower at the greater signal-masker frequency separation. It is suggested that the results for the two maskers with the greatest bandwidth (W and V) were influenced by combination bands produced by the interaction of components within the masker. The results for the maskers with very small bandwidths (S, N, and F), suggest that the upper side of the auditory filter increases in slope with increasing level.
使用频率始终高于信号频率的掩蔽音，获得了三个频率（fs）分别为0.25、1.0和4.0千赫兹的正弦信号的掩蔽增长函数。使用了五种不同的掩蔽音，其选择目的是评估掩蔽音中的时间波动以及掩蔽音内各成分相互作用产生的组合产物的影响：一个正弦波（S）；带宽为16赫兹的窄带噪声（N）；带宽略宽、等于每个fs处听觉滤波器等效矩形带宽（ERB）的0.75倍的噪声（W）；带宽非常宽、等于0.4fs且下限截止频率和频谱水平与掩蔽音W匹配的噪声（V）；以及由噪声调制的正弦载波频率，且带宽和中心频率与掩蔽音N匹配的（F）。掩蔽音S、N、W和F的中心频率要么是1.1fs，要么是1.2fs。掩蔽音S通常产生的掩蔽量最小，其掩蔽增长函数的斜率最浅（远小于1）。掩蔽音N和F的结果类似；两者产生的掩蔽都比掩蔽音S略多，且掩蔽增长函数的斜率比掩蔽音S略大。掩蔽音W和V产生的掩蔽比其他掩蔽音更多，且掩蔽增长函数的斜率更陡。对于所有掩蔽音，掩蔽增长函数的斜率在信号 - 掩蔽音频率间隔较大时较低。有人认为，带宽最大的两种掩蔽音（W和V）的结果受掩蔽音内各成分相互作用产生的组合频带影响。带宽非常小的掩蔽音（S、N和F）的结果表明，听觉滤波器的上侧斜率随声级增加而增大。
