The Development of Low-Level Speaker-Specific Information for Speaker Recognition

是什么帮助我们仅仅根据人们的言语来识别他们？语音信号中有没有属性可以帮助我们识别人们，即使他们试图隐藏自己的身份？在国家科学基金会的支持下，卡罗尔·埃斯皮尔-威尔逊博士将进行研究，以回答这些问题，并开发自动算法，从语音信号中可靠地提取特定于说话人的信息。这项研究将把来自言语产生和言语感知的知识结合起来。该项目的更广泛影响包括应用于各种语言障碍的诊断工具、开发语音训练辅助工具、开发个性化的人机交互、语音识别以及加深对发音-声学映射的理解。该项目通过进一步发展马里兰大学帕克分校的演讲相关课程，并为本科生和研究生提供参与研究的机会，将研究和教育结合在一起。

What helps us to recognize people on the basis of their speech alone? Are there properties in the speech signal that will help us recognize people even if they are trying to conceal their identity? With support from the National Science Foundation, Dr. Carol Espy-Wilson will conduct research to answer these questions and to develop automatic algorithms that will reliably extract speaker-specific information from the speech signal. This research will bring together knowledge from speech production and speech perception. The broader impacts of this project include applications to diagnostic tools for various speech disorders, the development of speech training aids, the development of personalized human-computer interaction, speech recognition, and a deeper understanding of the articulatory-to-acoustic mapping. This project integrates research and education by furthering the development of speech-related courses at the University of Maryland at College Park, and providing opportunities for undergraduate and graduate students to participate in research.