A psychoacoustic approach to dysphonic voice quality perception in vowels

描述(申请人提供)：嗓音质量障碍是语音障碍患者最常见的问题之一。了解发声障碍嗓音质量的声学基础并开发其测量工具是评估和治疗这些患者的关键步骤。这些变化往往是治疗结果的主要衡量标准。了解声带生理学、声学和声质感知之间的关系对于指导这些患者的治疗策略也是必要的。虽然语音质量的研究在过去的几十年里引起了极大的关注，但它也受到了一些问题的困扰。主观方法量化发声质量是高度不一致的，缺乏有效性。客观方法灵敏度、特异度有限，缺乏一致性。因此，不存在被广泛接受的或标准化的语音质量评估格式。我们之前的研究已经确定了导致这些问题的几个原因。这些问题包括：(1)控制不充分的知觉实验，(2)声刺激及其知觉关联之间的非线性映射，以及(3)每种声音质量的多个声学线索的存在(通常显示交易关系)。我们已经能够通过使用心理测量学原理重新设计知觉实验，并通过使用听觉处理模型作为信号处理前端来解释这些因素。实际上，这使我们能够研究声学信号，因为它是在听觉系统中表示的。这种声音信号的“内部”表示被发现提供了呼吸声音质量的测量，比用于相同目的的传统测量高达25%。拟议的研究是这项工作的延伸。我们寻求(A)进一步改进获得准确的语音质量感知判断的方法，(B)使用这些方法来确定粗糙和应变的声学线索，(C)开发计算模型来解释发声障碍语音质量的感知，以及(D)评估这些模型预测语音质量感知判断的成功。然后，这些发现将被转化为临床实践，因为这些可以指导治疗策略，并导致制定评估治疗结果的措施。

DESCRIPTION (provided by applicant): Disorders of voice quality are one of the most frequent problems seen in patients with voice disorders. Understanding the acoustic bases for dysphonic voice quality and developing tool for its measurement is a critical step in the assessment and treatment of these patients. These changes are often the primary measure of treatment outcome. Knowledge of the relationships between vocal fold physiology, acoustic and the perception of vocal quality are also necessary to guide treatment strategies for these patients. Although the study of voice quality as attracted immense attention in the past several decades, it has been plagued by a number of problems. Subjective approaches to quantify vocal quality are highly inconsistent and lack validity. Objective methods have limited sensitivity, specificity and lack consistency. Therefore, no well-accepted or standardized format for the evaluation of voice quality exists. Our previous research has identified several reasons that contribute to these problems. These include: (1) inadequately controlled perceptual experiments, (2) non-linear mapping between the acoustic stimulus and its perceptual correlate, and, (3) the presence of multiple acoustic cues for each voice quality (which often show trading relationships). We have been able to account for these factors by redesigning perceptual experiments using psychometric principles and through the use of an auditory processing model as a signal processing front-end. In effect, this allows us to study the vocal acoustic signal as it is represented in the auditory system. This "internal" representation of the vocal signal was found to provide measures of breathy voice quality that were as much as 25% better than conventional measures for the same purpose. The proposed research is an extension of this work. We seek to (a) further refine the methods for obtaining precise perceptual judgments of voice quality, (b) use these methods to determine the acoustic cues for roughness and strain, (c) develop computational models to explain the perception of dysphonic voice quality and, (d) evaluate the success of these models to predict the perceptual judgments of voice quality. The findings will then be translated into clinical practice as these can guide treatment strategies and lead to the development of measures to assess treatment outcome.