摘 要在现在互联网霸屏的时代，语音的识别功能可当成完成人和电脑相互交换的 热门技术，有待深究。当今社会有关 deep learning 能到大量数据库里面查找重要 信息，而 deep learning 就是模式识别范畴的热点。以它当做理论基础，对语音 识别很有现实意义，deep learning 主要是采用各个层次的非线性变换的提取信息 技能，因为这个系统含有层次化的特色，可以用这个方式去建立模型。这篇文章 在内容上给出了语音识别以及 deep learning 的基本知识，重点介绍怎样把 deep learning 应用到语音识别的研究里面。69253

面对目前智能计算机及大规模数据的发展，依据大脑处理语音、图像数据方 法的 deep learning 技术应运而生。传统的语音识别技术对特征筛选的人工技能 要求高，而且准确率低 deep learning 技术是应用在音频信号识别、模仿大脑的语 音信号学习以及识别的模式。在音频信号处理时，运用 deep learning 进行音频 数据的特征提取以及培训，会大幅度提高音频信号识别的准确性。

该毕业论文有图 10 幅，表 2 幅，参考文献 27 篇。

关键词: deep learning 语音识别 MATLAB 信号处理

The Research of Speech Recognition Based on Deep Learning

Abstract

In now the era of Internet screen, voice recognition function can be as complete and the computer exchange of hot technology, to be the bottom. In today's society, deep learning can find important information in a large number of databases, and learning deep is a hot spot in pattern recognition field. To it as a theoretical basis, has great practical significance to speech recognition, deep learning is mainly the nonlinear transform of the various levels of information extraction skills, through the level of, the realization of the modeling of the program. This paper mainly introduces the basic knowledge of speech recognition and learning deep, and focuses on how to apply learning deep to speech recognition research.

In view of development of computers and bid, data, the technology of deep learning, on the basis of voice and image processing, come into being,. Traditional technology of speech sounds demands high quality of personal skills, and it is a cc u r a c v is lower, applying deep learning to the recognition of speech sounds,imitating, the speech learning, and recognition of the brain.Utilizing, deep learning, to filter and train the features, during, the process of voice analysis, will rise the accuracy of the recognition of speech massively.

Key Words: deep learning recognition of speech MATLAB signal processing

目 录

摘 要 II

Abstract III

目 录 IV

图清单 VI

表清单 VI

1 绪论 1

1.1 Deep Learning 对语音识别的意义 1

1.2 语音识别的研究现状 1

1.3 研究问题及内容 1

1.4 论文结构 3

2 语音识别基础理论 4

2.1 语音识别基本原理 4

2.2 语音特征提取 4

2.3 语音模型 5

2.4 解码技术