【作者】 李璐；

【导师】 张笑燕；

【作者基本信息】 北京邮电大学 ， 软件工程， 2010， 硕士

【摘要】 语音识别技术的发展使得人与计算机的交互成为可能,针对目前汉语拼音在自学过程中缺乏反馈性意见的指导,本文把HSK大纲的初级词汇作为识别内容,结合语音识别的相关原理,利用VC7.0集成开发环境设计了汉语拼音的学习系统,系统按照学习、练习和测试三部分功能来实现,并在练习功能中嵌入语音识别技术与用户进行交互,给出用户发音的评价和反馈信息。在课程的组织中,利用ACCESS2000数据表的方式对课程内容进行存储和分类,利用DAO数据库驱动方式建立于Visual C++应用程序的数据交换,利用Visual C++7.0构建了学习系统的框架和功能的详细实现,并对用户信息和学习过程的存储和管理。在学习和测试功能中,结合FLASH和WAVE等多媒体技术来实现对拼音发音过程中的动态演示。文中Flash的播放通过利用第三方Shockwave控件完成,WAVE文件的播放通过Windows MCI API函数来实现。在练习环节中,通过DirectSound技术对用户发音的录入,存储为WAVE格式文件并利用SAPI5.1所提供的COM接口和语法规则来进行音节和声调的识别。本文介绍了识别中的动态时间弯折算法、语料库的建立、声韵分割技术以及评价分级标准,描述了系统识别系统的结构、功能和流程。最后给出了系统各个界面的功能说明和测试。更多还原

【Abstract】 With the development of speech recognition, the interactive between people and computer became possible. Aiming at the condition of Pinyin self-learning lack of tutor’s feedback suggestion, the Chinese pinyin learning system with speech recognition principle was designed using the tools of visual C++7.0 with the elementary content of HSK outline in this paper. The system included the three parts of follow me, practice and assignment, and in the practice module, the speech recognition technology was embedded and give out the evaluation and feedback information of user’s pronunciations.In the course organization, using access 2000 database to save and classify the course and setting up the data exchanging between the visual C++and database via the DAO database drivers. The frame of learning system and function of detail implementation was built, and the user’s basic information and learning process also were saved and managed.In the modules of follow me and assignments, the course of pingyin pronunciation was danaymic demonstrated by multimedia of flash and wave techniques. The flash playing was implemented by the third party tool of shockwave control and wave file playing done by Windows MCI API functions.In the module of practice, throught the DirectSound technology, the user’s pronunciation speech was recorded and saved as wave format file. Using the interface of COM provided by SAPI5.1 and grammar rules, the recorded file’s syllable and tone were recognized. The arithmetic of dynamic time warping, speech material building, and segmentation of initial and final part and criterion of pronunciation grade were introduced in the process of recognization. The structure, function and flow of recognization system were described. And the user interface of application software was narrated and tested.更多还原