面向情感变化检测的语音情感数据库  

Emotional Change Detection Oriented Speech Emotion Database

作　　者：张会云 黄鹤鸣[1,2,3] 李伟 ZHANG Hui-yun;HUANG He-ming;LI Wei(School of Computer Science,Qinghai Normal University,Xining Qinghai 810008,China;Key laboratory of Tibetan information processing,Ministry of education,Xining,Qinghai 810008,China;Tibetan information processing and Machine Translation Key Laboratory of Qinghai Province,Xining,Qinghai 810008,China)

机构地区：[1]青海师范大学计算机学院,青海西宁810008 [2]藏文信息处理教育部重点实验室,青海西宁810008 [3]青海省藏文信息处理与机器翻译重点实验室,青海西宁810008

出　　处：《计算机仿真》2021年第9期448-455,共8页Computer Simulation

基　　金：青海省自然科学基金(2016-ZJ-904);国家自然科学基金(61662062,61462072)。

摘　　要：语音情感数据库是进行语音情感识别研究的基础。多样化、大规模、高质量的优质的语音情感数据库对语音情感识别系统性能的提升具有重要作用。研究首先全面地分析了目前国内外使用最广泛的语音情感数据库,并将其归纳为单模态语音情感数据库和多模态语音情感数据库;然后通过实验将不同数据库上所取得的最优性能进行对比,选出可推荐的优质语音情感数据库,为相关研究人员提供有价值的学术参考;最后对未来语音情感数据库的建立提出几点建议。Speech emotion database is the basis of speech emotion recognition.Diversified, large-scale and high-quality speech emotion database plays an important role in improving the performance of speech emotion recognition system.Firstly, the most widely used speech emotion database at home and abroad was comprehensively analyzed and classified into single mode speech emotion database and multi-mode speech emotion database.Then, through the experiment, the optimal performance of different databases was compared, and the recommended high-quality voice emotion database was selected to provide valuable academic reference for relevant researchers.Finally, some suggestions were put forward for the establishment of future speech emotion database.

关 键 词：语音情感识别 语音可懂度 人机交互 语音情感数据库 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]