Department of Audiology and Speech Pathology, University of Tennessee, Health Sciences Center, Knoxville, Tennessee 37996, USA.
J Voice. 2012 Nov;26(6):817.e1-13. doi: 10.1016/j.jvoice.2012.04.011. Epub 2012 Aug 24.
OBJECTIVE/HYPOTHESIS: This study investigated whether listener's experience, presence/absence of vibrato, formant frequency difference, or onset delay affect the ability of experienced and inexperienced listeners to segregate complex vocal stimuli.
Repeated measures factorial design.
Two sets of stimuli were constructed: one with no vibrato and another with vibrato. For each set, each stimulus was synthesized at four pitches: A3, E4, B4, and F5. Stimuli were synthesized using formant patterns appropriate for the vowel |ɑ|. Frequencies for formants one through four were systematically varied from lower to higher in an attempt to simulate the acoustic results of corresponding changes in vocal tract length. Four formant patterns were synthesized (patterns A-D). Three pairs were created at each pitch, pairing the formants AB (mezzo-soprano/mezzo-soprano), CD (soprano/soprano), and AD (mezzo-soprano/soprano). Each of these three pairs was constructed in three separate conditions: simultaneous onset; the first voice in the pair with an onset delay of 100 milliseconds; and the second voice in the pair with an onset delay of 100 milliseconds. Using a scroll bar, listeners rated how difficult it was for them to hear each stimulus pair as two separate voices.
The most difficult combinations to segregate were produced with no vibrato and used simultaneous onset. The easiest conditions to segregate were combinations including a "soprano-like" formant pattern (D) in the vibrato condition. Overall, listener's experience did not affect the perceived difficulty of segregation; however, in the presence of vibrato cues, inexperienced listeners did not use delay cues as an aid in segregation in the same manner as did experienced listeners. Once vibrato was removed from the experimental context, inexperienced listeners were able to use delay to aid in segregation in a similar manner to experienced listeners.
Presence/absence of vibrato, formant pattern difference, and onset delay interact in a complex manner to affect the perceived difficulty of voice segregation.
目的/假设：本研究旨在探讨听众经验、颤音的有无、共振峰频率差异或起始延迟是否会影响有经验和无经验听众分离复杂声音刺激的能力。
重复测量因子设计。
构建了两组刺激：一组无颤音，另一组有颤音。对于每组，每个刺激都以四个音高进行合成：A3、E4、B4 和 F5。使用适合元音|ɑ|的共振峰模式合成刺激。共振峰一到四的频率从低到高系统地变化，试图模拟声道长度相应变化的声学结果。合成了四个共振峰模式（模式 A-D）。在每个音高下，每个模式都创建了三个对，将共振峰 AB（女中音/女中音）、CD（女高音/女高音）和 AD（女中音/女高音）配对。这三个对中的每一个都在三个单独的条件下构建：同时起始；对中的第一个声音有 100 毫秒的起始延迟；对中的第二个声音有 100 毫秒的起始延迟。听众使用滚动条对每个对的两个单独声音的可听度进行评分。
最难分离的组合是在无颤音且使用同时起始的条件下产生的。最容易分离的条件是在颤音条件下包含“女高音样”共振峰模式（D）的组合。总的来说，听众经验并没有影响到感知分离的难度；然而，在有颤音提示的情况下，无经验的听众并没有像有经验的听众那样，将延迟提示用作分离的辅助手段。一旦从实验环境中去除颤音，无经验的听众就能够以类似于有经验听众的方式使用延迟来辅助分离。
颤音的有无、共振峰模式差异和起始延迟以复杂的方式相互作用，影响声音分离的感知难度。
