EI / SCOPUS / CSCD 收录

中文核心期刊

汉语语音资料库的语音学标记及人工切分

Manual labeling in Chinese speech database

  • 摘要: 介绍了汉语语音综合资料库的一个子库:CAS-SYL。该数据库包括汉语全部有调音节1267个,共计10个发音人;全部语音数据由人工完成音段切分及语音学标注。针对汉语音节的声韵结构,语音学标注水平被定位在半音节层次上.语音学标注符号系统采用了计算机可读的音标符号系统一汉语SAMPA-X(extended SAM Phonetic Alphabet).还介绍了语音学标注策略,音段定位原则,基于语音波形的声门关闭时刻:GCI(Glottal Closed Instant)的声学线索。同时对声韵间的协同发音的声学体现进行了总结。最后对人工切分带来的非稳定性进行了分析.

     

    Abstract: In this paper we present that CAS--SYL, a subeorpora of the CAS Chinese Speech Database,which includes total l,267 Chinese isolated syllables read by 10 speakers, has been labeled by a well trained transcriber. Labeling level is at semi--Syllable level, it's mainly due to the initial/final construction of Chinese syllable. The phonetic symbol system adopted to annotate the utterance segment is SAMPA--X for Chinese. Placement of segment boundary and labeling strategies are presented. We also show the clues of GCI implicating in speech waveform that can used to refine the boundary of segment. The acoustical realization of crvarticulation happening in the transition from initial to final is summed also.And the factors resulting of the inconsistency of manual segmentation are discussed.

     

/

返回文章
返回