深度视觉语音生成研究进展与展望  

Reasearch progress and prospects of deep learning for visual speech generation

作　　者：刘丽 隋金坪 丁丁 赵凌君 匡纲要[1] 盛常冲 LIU Li;SUI Jinping;DING Ding;ZHAO Lingjun;KUANG Gangyao;SHENG Changchong(College of Electronic Science and Technology,National University of Defense Technology,Changsha 410073,China;Operational Software and Simulation Institute,Dalian Navy Academy,Dalian 116016,China;Center for Teaching and Research Support,National University of Defense Technology,Changsha 410073,China;National Key Laboratory of Electromagnetic Energy,Naval University of Engineering,Wuhan 430033,China)

机构地区：[1]国防科技大学电子科学学院,湖南长沙410073 [2]海军大连舰艇学院作战软件与仿真研究所,辽宁大连116016 [3]国防科技大学教研保障中心,湖南长沙410073 [4]海军工程大学电磁能技术全国重点实验室,湖北武汉430033

出　　处：《国防科技大学学报》2024年第2期123-138,共16页Journal of National University of Defense Technology

基　　金：国家自然科学基金资助项目(61872379)。

摘　　要：为了进一步推进深度学习技术驱动的视觉语音生成相关科学问题的研究进展,阐述了视觉语音生成的研究意义与基本定义,并深入剖析了该领域面临的难点与挑战;在此基础上,介绍了目前视觉语音生成研究的现状与发展水平,基于生成框架的区别对近期主流方法进行了梳理、归类和评述;最后探讨视觉语音生成研究潜在的问题和可能的研究方向。In order to further advance the development of visual speech learning,the task definition and research significance of visual speech generation was expounded and the difficulties and challenges were deeply analyzed in this field.Besides,the current status and development level of visual speech generation research was introduced,and the recent mainstream methods were sorted,classified and commented based on the difference of generation frameworks.At the end of the paper,the potential problems and possible research directions of visual speech generation were discussed.

关 键 词：视觉语音生成 深度学习 计算机视觉 计算机图形学 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]