【作者】 邓杏娟；

【导师】 陈骥；

【作者基本信息】 重庆大学 ， 生物医学工程， 2008， 硕士

【摘要】 失语症患者的康复治疗问题越来越引起人们的关注。传统的失语症诊断及康复治疗都需要治疗师的参与,特别是在言语方面的诊断及训练更需要治疗师的主观判断,致使评估结果缺乏客观的定量和定性指标。随着计算机科学技术的发展,特别是语音识别技术的日趋成熟,设计一套智能化、小型化、可灵活扩展的失语症辅助诊断和康复治疗为一体的系统软件是失语症康复治疗研究的一个重要方向。本课题主要论述了失语症辅助诊断及康复治疗系统的分析、设计及建立。重点论述基于语音识别技术和人工神经网络的失语症语音评估系统的实现。通过对系统需求的分析,设计出合理的数据库。对目前流行软件开发语言的分析,确定系统的开发方式。最后,通过图片及表格的方式来显示系统建立的情况。本课题主要的研究工作包括:关于语音信号的端点检测及声韵母的切分算法、基于隐马尔可夫模型(HMM)和BP神经网络的失语症语音评估系统的研究、语音信号信息量和流利性的评估及人工智能调度功能的设计思想。在这些研究的基础上,最后构建出失语症辅助诊断及康复治疗系统,并实现了病历信息子系统、失语症辅助诊断子系统、康复治疗子系统及训练资料库子系统之间的联系。更多还原

【Abstract】 Aphasia rehabilitation treatment problems have received much attention. The methods of traditional aphasia diagnosis and rehabilitation are accomplished with the help of therapists. The diagnosis and treatment mainly depend on the subjective judgment of therapists, which is lack of an objective assessment of quantitative and qualitative indicators, especially in the area of speech training. As the fast development of computer and the technology mature in speech recognition, designing a set of system software that is intelligent, small size, and scalable for the aphasia diagnosis and rehabilitation treatment is becoming one of the important directions of aphasia rehabilitation.The main contents described in the dissertation are the diagnosis and rehabilitation treatment system of aphasia about analysis, design and implementation. It is focus on the aphasia speech evaluation system based on speech recognition technology and artificial neural network. The reasonable database is devised by analyzing the system’s demands and the development approach of system is decided through the analysis of current software development language. At the end, the achievement of system that is designed by the author is showed in the ways of pictures and tables.The major research in the dissertation is summarized as follows:1) Proposed a new and reasonable scheme on the algorithm of the endpoint detection and segmentation between consonants and vowels based on the system’s demands; 2) Studied hard on the aphasia speech evaluation system based on Hidden Markov Model (HMM) and BP neural network and the assessment of speech signal fluency and information; 3) Introduced a new idea of artificial intelligence and scheduling function in the designed system; 4) On the basis of those, constituted the diagnosis and rehabilitation treatment system of aphasia, and made the medical information subsystem, the diagnosis of aphasia subsystems, the rehabilitation subsystem and the training database subsystem come true .更多还原