On The Use of Non-Verbal Speech Sounds in Human Communication

本课题研究的重点是“对话结构”与“积极倾听”过程之间的关系。该项目没有产生具体的新技术，但让我们更好地了解了口语的本质，这将使我们能够改进语音合成和识别技术。我们的研究主题启发了新的欧洲社会信号处理项目。此外，我们还收集了与三个国家的其他大学合作的大型多模式语料库，并将数据库发布在(http://www.speech-data.jp/nick/mmx/d64.html).网站上最后，我们利用本研究开发的图像和语音处理模块，制作了一个机器人Herme的原型。该机器人目前正在爱尔兰展出，它正在收集一个新的互动语音语料库。

This project focused on the relationships between "dialogue structure" and the processes of "active listening". The project did not result in specific new technology but gave us a better understanding of the nature of spoken language in a way that will allow us improve speech synthesis and recognition technology. Our research theme inspired the new European project on Social Signal Processing. Further- more, we collected a large multi-modal corpora cooperating with other universities in three countries, published the database on the website (http://www.speech-data.jp/nick/mmx/d64.html). We finally made a prototype of a robot "Herme" which employs image and speech processing modules developed in this study. The robot is currently in exhibition in Ireland where it is collecting a new corpus of interactive speech.