Richards Virginia M, Tisby Mariel Kazuko, Suzuki-Gill Eli N, Shen Yi
Department of Cognitive Sciences, University of California, Irvine, California 92687, USA.
Department of Speech and Hearing Sciences, University of Washington, Seattle, Washington 98105, USA.
J Acoust Soc Am. 2021 Mar;149(3):1567. doi: 10.1121/10.0003646.
When spectral components of a complex sound are presented not simultaneously but distributed over time, human listeners can still, to a degree, perceptually recover the spectral profile of the sound. This capability of integrating spectral information over time was investigated using a cued informational masking paradigm. Listeners detected a 1-kHz pure tone in a simultaneous masker composed of six random-frequency tones drawn on every trial. The spectral profile of the masker was cued using a precursor sound that consisted of a sequence of 50-ms bursts, separated by inter-burst intervals of 100 ms. Each burst in the precursor consisted of pure tones at the masker frequencies with tones appearing at each of the masker frequencies at different presentation probabilities. As the presentation probability increased in different conditions, the detectability of the target improved, indicating reliable precursor cuing regarding the spectral content of the masker. For many listeners, performance did not significantly improve as the number of precursor bursts increased from 2 to 16, indicating inefficient integration of information beyond 2 bursts. Additional analyses suggest that when intensity of the bursts is relatively constant, the contribution of the precursor is dominated by information in the initial burst.
当复杂声音的频谱成分不是同时呈现而是随时间分布时，人类听众在一定程度上仍能从感知上恢复声音的频谱轮廓。使用线索化信息掩蔽范式研究了这种随时间整合频谱信息的能力。听众在每次试验中由六个随机频率音调组成的同时掩蔽中检测到1kHz纯音。掩蔽的频谱轮廓通过一个前导声音进行提示，该前导声音由一系列50毫秒的脉冲组成，脉冲间间隔为100毫秒。前导中的每个脉冲由掩蔽频率的纯音组成，每个掩蔽频率的音调以不同的呈现概率出现。随着在不同条件下呈现概率的增加，目标的可检测性提高，表明关于掩蔽频谱内容的前导提示是可靠的。对于许多听众来说，当前导脉冲的数量从2增加到16时，表现并没有显著提高，表明超过2个脉冲后信息整合效率低下。进一步分析表明，当脉冲强度相对恒定时，前导的贡献主要由初始脉冲中的信息主导。
