藏语语音识别研究进展综述  

An Overview of the Research Progress of Tibetan Speech Recognition

作　　者：边巴旺堆 王希 王君堡 Bianba-Wangdui;WANG Xi;WANG Junbao(School of Information Science and Technology,Tibet University,Lhasa 850000,China;Information Technology National Experiment Teaching Demonstration Center,Tibet University,Lhasa 850000,China)

机构地区：[1]西藏大学信息科学技术学院,西藏拉萨850000 [2]信息技术国家级实验教学示范中心,西藏拉萨850000

出　　处：《高原科学研究》2022年第4期76-84,共9页Plateau Science Research

基　　金：西藏自治区高原通信科研创新团队项目(XZJY2018003)。

摘　　要：语音识别研究内容主要包括孤立词识别、连续语音识别、语音情感识别以及语音数据库构建等。文章首先从识别模型、识别率、评价指标等方面对藏语孤立词和连续语音识别研究进行了文献回顾与评述,并对近几年才开始开展的藏语语音情感识别研究现状进行了分析;然后从研究者自用和开放使用两种情况来介绍藏语语音数据库的构建、使用情况;最后分析了目前藏语语音识别研究的主要难点,并对可能的发展方向进行了展望。Tibetan speech recognition study mainly includes isolated word recognition,continuous speech recognition,speech emotion recognition,speech database construction,etc.In this paper,firstly,the study of the Tibetan isolated words and continuous speech recognition is introduced from several aspects,such as the recognition model,recognition rate,and evaluation index,and the overview of Tibetan speech emotion recognition which has been studied in recent years is given.Then,from the situation of self-use and open-use,the construction and usage of the Tibetan speech database are introduced.Finally,the main difficulties in the study of Tibetan speech recognition are discussed,and possible development directions are proposed.

关 键 词：藏语 语音数据库 语音识别 语音情感识别 

分 类 号：TN912.34[电子电信—通信与信息系统]