Centre for Neuroscience, University of Melbourne, Melbourne, Australia.
J Voice. 2011 Mar;25(2):137-49. doi: 10.1016/j.jvoice.2009.09.003. Epub 2010 Feb 19.
Assessment of the voice for supporting classifications of central nervous system (CNS) impairment requires a different practical, methodological, and statistical framework compared with assessment of the voice to guide decisions about change in the CNS. In experimental terms, an understanding of the stability and sensitivity to change of an assessment protocol is required to guide decisions about CNS change. Five experiments (N = 70) were conducted using a set of commonly used stimuli (eg, sustained vowel, reading, extemporaneous speech) and easily acquired measures (eg, f₀-f₄, percent pause). Stability of these measures was examined through their repeated application in healthy adults over brief and intermediate retest intervals (ie, 30 seconds, 2 hours, and 1 week). Those measures found to be stable were then challenged using an experimental model that reliably changes voice acoustic properties (ie, the Lombard effect). Finally, adults with an established CNS-related motor speech disorder (dysarthria) were compared with healthy controls. Of the 61 acoustic variables studied, 36 showed good stability over all three stability experiments (eg, number of pauses, total speech time, speech rate, f₀-f₄. Of the measures with good stability, a number of frequency measures showed a change in response to increased vocal effort resulting from the Lombard effect challenge. Furthermore, several timing measures significantly separated the control and motor speech impairment groups. Measures with high levels of stability within healthy adults, and those that show sensitivity to change and impairment may prove effective for monitoring changes in CNS functioning.
评估中枢神经系统（CNS）损伤的语音需要与评估语音以指导 CNS 变化决策不同的实用、方法学和统计框架。从实验角度来看，需要了解评估方案的稳定性和对变化的敏感性，以指导关于 CNS 变化的决策。进行了五项实验（N=70），使用了一组常用的刺激（例如，持续元音、朗读、即兴演讲）和易于获得的测量方法（例如，f₀-f₄、停顿百分比）。通过在健康成年人中在短时间和中间重测间隔（即 30 秒、2 小时和 1 周）内重复应用这些措施来检查这些措施的稳定性。然后，使用可靠地改变语音声学特性的实验模型（即 Lombard 效应）来挑战那些被认为稳定的措施。最后，将患有已确立的与 CNS 相关的运动言语障碍（构音障碍）的成年人与健康对照组进行比较。在所研究的 61 个声学变量中，有 36 个在所有三个稳定性实验中表现出良好的稳定性（例如，停顿次数、总言语时间、言语速度、f₀-f₄）。在具有良好稳定性的测量中，一些频率测量值在响应 Lombard 效应挑战导致的声音努力增加时发生了变化。此外，一些计时测量值显著区分了对照组和运动言语障碍组。在健康成年人中具有高稳定性的测量值，以及那些表现出对变化和损伤的敏感性的测量值，可能对监测 CNS 功能变化有效。
