基于依存句法分析的汉语韵律层级自动预测技术研究  

A Study on Chinese Prosodic Hierarchy Prediction Based on Dependency Grammar Analysis

作　　者：邵艳秋[1] 穗志方[1] 韩纪庆[2] 吴云芳[1] 

机构地区：[1]北京大学计算语言学研究所,北京100871 [2]哈尔滨工业大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《中文信息学报》2008年第2期116-123,共8页Journal of Chinese Information Processing

基　　金：国家自然科学基金资助项目(60503071);国家重点基础研究发展规划973资助项目(2004CB318102)

摘　　要：不同的韵律层级可以将文本划分成适合朗读与理解的韵律组块,从而保证合成语音能够以自然的节奏表现出来。目前对韵律层级预测所采用的特征绝大多数是较为浅层的特征,如词性、词长等,但这些浅层特征对有的韵律层次如韵律短语的预测能力比较弱。实际上,句法结构同韵律层级之间有着非常紧密的联系,二者相互影响,相互制约。本文根据依存句法分析的结果,抽取出若干同韵律层级相关的深层句法特征对韵律层级进行预测。实验证明,其中内弧跨度和内弧类型等特征,对浅层特征较难解决的类似韵律短语这种中间层次的韵律单元划分问题,可以起到很大的提高作用,使韵律短语标注的综合F值提高了11%。Different prosodic hierarchy could divide texts into several prosodic chunks for better speaking and understanding. Currently, many shallow features such as part-of-speech, length of word are used to predict the prosodic hierarchy. But these features are not powerful for some prosodic unit prediction such as prosodic phrase. In fact, syntactic structure is in close touch with prosody structure. They influence and restrict each other. In this paper, based on dependency grammar, some deep features which are related with prosody hierarchy are extracted. Compared to the shallow features, the deep features such as inner-arc span and inner-arc type are more effective on the prediction of the middle level such as prosodic phrase. The F-score increases about 11%.

关 键 词：计算机应用 中文信息处理 语音合成 韵律层级 句法结构 依存分析 停顿指教 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]