基于句子级的朝鲜语口语语音库的研究与建立

发布时间：2018-02-04 12:33

本文关键词： 朝鲜语口语语音库词频统计　出处：《科技通报》2017年04期 　论文类型：期刊论文

【摘要】：各个少数民族语言语音库的建立,为语言研究奠定了坚实的基础。但延边地区朝鲜语口语的语音库建设目前还处于空白状态。为使语音库的内容更具有普遍意义与应用意义,本文利用PHP强大的分词库统计出大量朝鲜语文章中各音节或词语的出现频率,以此词频统计结果为依据组建朝鲜语常用短句,使短句中各音节或词语出现的概率尽量接近上述统计结果。其次,以组建的常用句为内容,召集朝鲜族男、女各5名进行录音,切分后的7200个短句用数据库管理,建立了基于句子级的延边地区口语常见句的语音库。最后采用MATLAB软件与SQL server的接口测试,验证数据库管理语音材料的有效性。
[Abstract]:The establishment of each minority language corpus. It has laid a solid foundation for language research. However, the construction of the corpus of spoken Korean in Yanbian area is still in a blank state. In order to make the content of the corpus more universal and practical significance. This paper makes use of PHP's powerful thesaurus to calculate the occurrence frequency of every syllable or word in a large number of Korean articles. Based on the result of word frequency statistics, this paper constructs the commonly used short sentences in Korean. The probability of each syllable or word appearing in the short sentence is as close as possible to the above statistical results. Secondly, taking the commonly used sentences as the content, five Korean men and five women are called to record. The 7 200 short sentences are managed by database, and the common sentences in Yanbian area are built based on sentence level. Finally, the interface test between MATLAB software and SQL server is adopted. Verify the effectiveness of database management of voice materials.
【作者单位】：延边大学工学院;
【基金】：吉林省科研项目(项目编号:612016015)
【分类号】：H219;TP311.13
【正文快照】： 随着信息技术与计算机科学的发展,语音技术使机器行为与人类自然语言的交互成为可能,不论是语音合成、语音识别还是语音辨认研究,都必定依靠于后端优秀语音语料库的建设[1]。早在1970年,英语口语语料库就在英国诞生[2]。目前国外语音库的发展较为成熟,中国的语音库研究也已在

【相似文献】