声带小结的声学特征研究

发布时间：2018-11-27 14:58

【摘要】：随着我国工业化程度的不断提高，，引起了喉部疾病发病率的不断攀升。同时，在那些需要长期使用嗓音的职业中，喉部疾病呈现一种常态职业病的倾向。目前喉部疾病最常用的检查方法是采用喉内窥镜进行检查，这种有创式接触式的检查方式，操作复杂，病人有不适甚至痛苦感，部分病人甚至畏惧不愿意接受检查，以致耽误了治疗时间。另外，内窥镜光纤管道直接进入人体喉部的检查方式，还存在设备消毒不彻底而引发交叉感染的风险。因此从改进喉部疾病的诊断方法，降低人民群众的医疗检查费用等方面来考虑，研究一种安全的、无创的、费用低的检查方法，显得尤为必要。此前的国内外病态嗓音识别工作和基于声学信号的喉部疾病诊断主要以健康嗓音与病态嗓音的分类识别为研究重点，其中大多数方法基于统计学或者模糊判别，无物理数学模型。这反映了目前的研究只注重于嗓音数据与信号本身，而忽略了信号的本质是对对象某些物理特性的反映。在这个过程中没有能够提出具有一定病理意义的指标，而已经提出的指标相对来说统一性差，实验重复性差异大，准确率低。本文将持续稳定的元音为研究对象，以人体声带的振动和发声原理为基础，将声带的振动简化为一般的理想自由弦振模型，结合波动方程的解和小结病变的物理性变化来推导声带小结病变后的声学信号特征，利用声带对称性和耦合性建立内禀标准来检测声带的小结病变，并给出了其频域的分析方法，提出差频作为声带小结的信号特征。在本课题的研究过程中，同时开发出了一套声学信号的采集、存储、分析系统。为了更好的收集、管理和分析嗓音信号，还专门编写了针对嗓音疾病的声学数据存储系统，建立了信号分析中常用的功能和一系列具有实用性的数字信号分析的图形化工具，使得整个系统更加专业化，具有针对性。实验结果表明多数情况下小结患者的声学信号的频谱的确会产生差频现象，利用这个特征能够在一定程度上区分正常嗓音组和病变嗓音组，这对于进一步研究声带小结的声学信号特征具有积极的意义。
[Abstract]:With the development of industrialization in China, the incidence of laryngeal diseases is rising. At the same time, laryngeal diseases tend to be normal in occupations that require long-term voice use. At present, the most commonly used examination method for laryngeal diseases is the use of laryngeal endoscopy, which is an innovative contact examination, which is complicated in operation, and patients have a sense of discomfort or even pain, and some patients are even afraid of being examined. To the point of delaying the treatment. In addition, there is a risk of cross-infection caused by incomplete disinfection of the equipment in the inspection mode of direct access to the larynx of the endoscope fiber-optic conduit. Therefore, it is necessary to study a safe, non-invasive and low cost examination method from the aspects of improving the diagnosis of laryngeal diseases and reducing the medical examination cost of the masses. Previous work on the recognition of pathological voice at home and abroad and the diagnosis of laryngeal diseases based on acoustic signals mainly focused on the classification and recognition of healthy voice and pathological voice, most of which were based on statistics or fuzzy discrimination. No physical mathematical model. This reflects that the present research only focuses on the voice data and the signal itself, but neglects that the essence of the signal is the reflection of some physical characteristics of the object. In this process, the indexes with certain pathological significance can not be put forward, but the indexes that have been proposed are relatively poor in unity, there are great differences in the repeatability of experiments, and the accuracy is low. In this paper, the continuous and stable vowels are taken as the object of study. Based on the vibration and sound principle of human vocal cords, the vibration of vocal cords is simplified as a general ideal free chord vibration model. Combined with the solution of wave equation and the physical changes of nodule lesions, the acoustic signal characteristics of vocal nodules were deduced, and the intrinsic standard was established to detect nodal lesions by using the symmetry and coupling of vocal cords. The analysis method in frequency domain is given, and the difference frequency is proposed as the signal feature of vocal cord summary. In the course of the research, a collection, storage and analysis system of acoustic signal is developed. In order to collect, manage and analyze voice signals better, a special acoustic data storage system for voice diseases is developed, and the commonly used functions in signal analysis and a series of practical graphical tools for digital signal analysis are established. Make the whole system more specialized, have pertinence. The experimental results show that in most cases, the spectrum of acoustic signals of nodule patients does produce differential frequency phenomenon, which can be used to distinguish the normal voice group from the diseased voice group to a certain extent. This is of great significance to the further study of acoustic signal characteristics of vocal nodules.
【学位授予单位】：重庆理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R767.41

【参考文献】