毕业论文-基于声学参数的病理嗓音细分研究.docVIP

毕业设计（论文） 基于声学参数的病理嗓音细分研究 中文摘要 对病理嗓音的细分识别研究，在临床医学和基础研究领域都有十分重要的意义。病理嗓音主要是由于声带和喉的各种疾病导致其闭合或振动异常，致使其声学性质发生改变，临床上表现出不同程度的声音嘶哑、失真等。结合病理嗓音的特点，各类疾病都可能引发喉部异常，导致嗓音的变化，传统的病理嗓音诊断方法都存在一些缺陷，因此需要采用声学检测的方法进行识别和处理。本课题挑选了125例甲状腺功能亢进、12例声带麻痹、24例胃液逆行、53例正常嗓音，主要需要研究常用的病理声学参数提取方法（基频微扰（Jitter）、幅度微扰（Shim）等），为病理嗓音细分识别做准备，并初步研究病理嗓音细分识别算法。采用weka软件进行识别实验进行分类，建模并提取算法交叉验证准确度达65%以上的结果，对实验结果进行分析讨论。 关键词：病理嗓音；weka；声学参数 pathological voice segmentation research based on acoustic parameters Abstract Subdivision of pathological voice recognition research, in the field of clinical and basic research has the very vital significance. Pathological voice is mainly due to the vocal cords and throat diseases led to its closure or abnormal vibration, its acoustic properties change, clinical showed different degrees of hoarseness, distortion, etc. Combining with the characteristics of pathological voice, all kinds of diseases may cause throat is unusual, lead to the change of the voice, the traditional pathological voice diagnosis methods have some defects, so the need to adopt the method of acoustic detection identification and processing. This topic selected 125 cases of thyroid function hyperfunction, 12 cases of vocal cord paralysis, 24 cases of retrograde gastric juice and 53 normal voice, the need to study the pathology of acoustic parameters commonly used extraction method (fundamental frequency perturbation (Jitter) and amplitude perturbation (Shim), etc.), preparing for pathological voice recognition segment, and a preliminary study of pathological voice recognition algorithm segment. Recognition experiments are classified with weka software, modeling and extracting algorithm of cross validation accuracy above 65% as a result, carries on the analysis discussion with the result of the experiment. Key words: pathological voice; Weka; The acoustic parameters 目 录 中文摘要 1 Abstract 2 第一章 引言 5 1.1 研究目的 5 1.2 近年国内外嗓音研究现状 5 1.3 本文的研究过程 6 第二章 病理嗓音的产生机理 7 2.1 语音的发音系统 7 2.2 语音的定义 7 2.3 语音生成数学模型 7 2.3.1 激励模型 8 2