摘要语音是人类交流的重要手段，在现实生活中语音情感识别技术具有重大的实用意义，这个技术的发展将对在生活中的各个领域带去变革性的影响，因此语音情感识别技术的实现富有重大意义，我们可以通过分析不同语音特征，并加以分析达到识别效果。25759
本文主要对于包含高兴，愤怒的语音以及正常语音进行特征提取并对比。本文提取的情感特征包括平均能量，能量变化率均值，平均基频，基频变化率均值，语速，第一共振峰均值，第一共振峰变化率均值，第一共振峰波动范围。由于原始语音不能直接提取特征参数，为此我们要对其进行加窗，分帧，预加重，请浊音判断等预处理。本文选取了通过判别短时能量，短时过零率，短时自相关函数这三个参数的基于双门限判别法的三参数清浊音判别法，并且运用自相关法计算基音频率，线性预测法估算共振峰频率。
本文的核心在于通过MATLAB编程提取上述参数，并结合PRAAT软件进行结果分析，综合MATLAB和PRAAT分析结果得出高兴，愤怒，平常状态下的语音特征，得出广泛规律。最后对于陌生语句采取特征提取并且进行对比分析，达到对于语音的情感识别效果。
关键词 情感特征 基音频率 共振峰 MATLAB编程 PRAAT分析
毕业论文设计说明书外文摘要
Title Emotional Recognition Technology of Speech
Abstract
Speech is the important means of human communication. Emotional Speech recognition technology has great practical significance in real life, the development of this technology will bring transformative effect in all areas of life. So the emotional recognition of speech technology has great significance. We can analyze different speech characteristics, and achieve recognition results.
This paper studies three kinds of emotions: happiness, anger and neural. Emotional characteristics in this paper include average energy extraction, energy rate mean, mean fundamental frequency, rate of change of the mean fundamental frequency, speed, the first formant mean, the mean rate of change of the first formant, the fluctuation range of the first formant. Prior to the emotional features extraction, original speech signal must be preprocessed. The process contains adding windows, sub-frame and pre-emphasis. The method we choose in the paper is three parameters criterion based on the traditional dual-threshold discrimination. The work of voicing judgments must be done before extracting the pitch. The three parameters consist of short-time energy, short-time zero-crossing rate, short-time autocorrelation function. And the approach of calculating the pitch in this paper is autocorrelation function. The method of estimating the first formant is linear prediction.
The core of this paper is to extract these parameters by MATLAB programming.And we use PRAAT to analyze the results. And we use the comprehensive analysis results to get the data of voice in happy, angry, and the usual state. Then we succeed to accomplish the work of emotional speech recognition.
Keywords emotion feature， the first format， pitch， MATLAB programming， PRAAT analysis
目 录 I
1 绪论 1
1.1 选题的背景及意义 1
1.2 国内外研究现状及面临的问题 2
1.3 本论文主要完成的工作 3
2情感语言基础 4
2.1 情感语音库 4
2.1.1 情感语音模型及语音库的分类 4
2.1.2 语音库原则及获取方法 6
2.2语音情感分析原理概述 7
2.3 语音信号预处理 7
2.3.1 语音信号的采集 7 MATLAB和PRAAT汉语普通话情感识别:http://www.youerw.com/tongxin/lunwen_19691.html