项目名称: 维吾尔语韵律结构的分析与预测模型的研究
项目编号: No.61462087
项目类型: 地区科学基金项目
立项/批准年度: 2015
项目学科: 自动化技术、计算机技术
项目作者: 姑丽加玛丽·麦麦提艾力
作者单位: 新疆师范大学
项目金额: 47万元
中文摘要: 韵律是言语的固有特征和自然属性,隐含了语法、语义、语用等多种语言功能。它不仅体现了说话人对文字的表达和理解,还反映说话人的态度、期望、情绪等高层信息。韵律在增加语言表达的自然度和可理解度、提高语音合成和语音识别系统的性能等方面扮演着非常重要的作用。项目在中英文的新理论、新方法和新技术的基础上,从维吾尔语声学、语法和词法特点出发,围绕韵律结构分析与预测所涉及的关键理论和技术,开展维吾尔语韵律结构分析和韵律结构标注语料库的建设,韵律特征参数分析(韵律特征参数选取、特征提取、统计分析),韵律结构预测(音节划分、分词、词干提取、词性预测、韵律特征参数预测等特征提取算法和统计建模方法),模型优化及性能分析等多个方面的研究,建立维吾尔语韵律结构分析与预测的理论体系及技术基础,并通过研发有关算法、工具和试验平台,在相关领域中应用与验证本课题取得的研究成果。
中文关键词: 维吾尔语;韵律结构;韵律特征参数;韵律边界预测
英文摘要: Prosody is most inherent and natural feature of language, implies syntatic, semantic, pragmatic functions. Not only reflects the expression and comprehension of speaker to the text, but the attitudes, expectations, emotions of speaker. Prosody plays more important role to increase naturalness and intelligibility of language expression, and to improve the performance of speech synthesis and speech recognition system. Considering the acoustic, syntactic and lexical features of Uyghur language, key technologies of prosodic structure analysis and prediction method are mainly studied in the project on the basis of new theory, method and technology of English and Chinese. The research including Uyghur prosodic structure analysis and construction of prosodic structure annotation corpus, prosodic feature analysis (feature selection, extraction, analysis), prosodic structure prediction(syllable and word segmentation,stem extracion,POS prediction,prosodic feature prediction and statistical model),optimization of prosodic model and performance analysis are conducted. System info and technological base of Uyghur prosodic structure analysis and prediction are built, and the research achievements of project are applied and verified in the relative area by developing related algorithms, tools and test platforms.
英文关键词: Uyghur language;prosodic structure;prosodic feature;prosodic boundary prediction