Chaos in human phonation and its measurement

描述(申请人提供)：嗓音障碍影响数百万美国人。嗓音障碍对生活质量的影响很大；当一个人的嗓音受损时，工作能力、与朋友和家人的关系、参加社会活动和参与日常生活的能力即使不是不可能，也会变得费力。声音产生的复杂性和声音障碍的非周期性决定了定量声学分析必须是非线性的。诸如抖动或抖动的线性参数依赖于线性信号，因此当应用于非周期性发音困难时是无效和不准确的。目前的非线性方法只适用于主要由声带振动(固有的有限阶)产生的非周期信号，而不适用于具有显著湍流(高阶)的信号。由于喘息引起的湍流气流是发音困难的一个共同特征，因此开发对带有突出喘息噪声的信号有效的声学分析方法是很重要的。我们将应用高阶非线性动力学理论对这些信号进行量化。我们还将进行计算机建模、切除喉部和人体实验，以确定产生湍流的机制。在项目1中，我们将优化能够分析具有显著湍流的语音信号的高阶非线性动力学参数。语音信号的传统描述使用三级系统，其中类型1是周期性的，类型2具有次谐波，类型3是非周期性的。将3型信号细分为主要由声带振动产生的信号和主要由湍流产生的信号将是有益的。目前的声学分析，无论是线性的还是非线性的，都不能分析主要由湍流产生的信号。高阶非线性参数如嵌入效率、广义维和广义熵可以量化这些信号。在项目2-3中，我们将使用计算机和切除的喉部模型进行实验，以阐明声门湍流的潜在机制。我们还将确定产生湍流能量所需的异常程度(例如息肉或声门间隙的大小)，从而使当前的非线性参数不再准确，必须使用高阶非线性分析。在项目4-5中，我们将评估人类发声的高阶特征。发音(元音、擦音)将会有所不同，以评估在声道中产生的湍流的影响。我们还将确定音量、发声类型(耳语、胸部、假声)和录音环境的影响。在良性肿块病变、水肿和声门功能不全的患者以及食道发声使用者中，将测量发声障碍的高阶非线性特征。最后，我们将确定高阶非线性参数相对于线性或当前非线性参数的灵敏度、特异度和可靠性。这五个项目结合了理论、基础科学、翻译和临床研究，以增强对无序发声的了解，并为临床医生提供对高维语音信号进行客观、定量的声学分析的有效方法。

DESCRIPTION (provided by applicant): Voice disorders affect millions of Americans. The impact of voice disorders on quality of life is significant; the ability to work, relate to friendsand family, participate in social activities, and engage in everyday life becomes effortful, if not impossible, when one's voice is impaired. The complexity of voice production and aperiodic nature of voice disorders necessitate that quantitative acoustic analysis be nonlinear. Linear parameters such as jitter or shimmer rely on a linear signal and thus are invalid and inaccurate when applied to aperiodic dysphonia. Current nonlinear methods are only valid for aperiodic signals created primarily by vocal fold vibration (which is inherently of limited order), not signas with prominent turbulent airflow (high order). As turbulent airflow due to breathiness is a common feature of dysphonia, it is important to develop methods of acoustic analysis which are valid for signals with prominent breathy noise. We will apply high-order nonlinear dynamic theory to quantify these signals. We will also perform computer modeling, excised larynx, and human subject experiments to determine the mechanisms by which turbulence is produced. In Project 1, we will optimize high-order nonlinear dynamic parameters capable of analyzing voice signals with prominent turbulent airflow. Traditional descriptions of voice signals use a three class system, with type 1 being periodic, type 2 having subharmonics, and type 3 being aperiodic. It would be beneficial to subdivide type 3 signals into those created predominantly by vocal fold vibration and those created predominantly by turbulent airflow. Current acoustic analyses, whether linear or nonlinear, are not capable of analyzing signals created predominantly by turbulent airflow. High-order nonlinear parameters such as embedding efficiency, generalized dimension, and generalized entropy can quantify these signals. In Projects 2-3, we will perform experiments using computer and excised larynx models to elucidate the mechanisms underlying glottal turbulence. We will also determine the level of abnormality required (e.g., size of polyp or glottal gap) required to produce turbulent energy such that current nonlinear parameters are no longer accurate and high-order nonlinear analysis must be employed. In Projects 4-5, we will evaluate the high-order characteristics of human phonation. Utterance (vowels, fricatives) will be varied to evaluate effects of turbulence created in the vocal tract. We will also determine the effects of volume, phonation type (whisper, chest, falsetto), and recording environment. High-order nonlinear characteristics of disordered voice production will be measured in patients with benign mass lesions, edema, and glottic insufficiency as well as esophageal voice users. Lastly, we will determine the sensitivity, specificity, and reliability of high-order nonlinear parameters compared to linear or current nonlinear parameters. The five projects combine theoretical, basic science, translational, and clinical research to enhance understanding of disordered voice production and provide clinicians with a valid method of performing objective, quantitative acoustic analysis on high-dimensional voice signals.