Herme, Yet Another Interactive Conversational Robot
Herme, Yet Another Interactive Conversational Robot
复制标题
DOI：
10.1109/acii.2013.127
复制
发表时间：
2013-09
期刊：
影响因子：
--
通讯作者：
J. Han;E. Gilmartin;N. Campbell
中科院分区：
文献类型：
--
作者：
J. Han;E. Gilmartin;N. Campbell
2012年，都柏林Trinity学院语音通信实验室开发的会说话的机器人Herme在三个月的时间里收集了近500份来自TCD科学画廊的临时访客的签名同意书。Herme是一个支持网络摄像头的小型乐高设备，并被编程为与任何进入她视野范围内的人交谈。她拍摄了互动过程，并在与她的新“朋友”聊天约三分钟后，试图说服他们签署一份伦理委员会同意书，允许她实验室的研究人员分析对话的音频视频记录，以便更好地了解未来人们如何与机器互动。该演示展示了在SFI资助下在语音通信实验室进行的FASTNET项目的工作，重点是社交谈话中的动作，将计算机语音合成与不同形式的语音和图像处理联系起来，以推断正在进行的对话是否成功。在科学画廊收集的对话录音语料库提供了人类社会互动如何遵循明确定义的模式，可以利用先进的语音技术的重要见解。在ACII 2013上展示的演示将使参与者能够与Herme互动并自行判断她的能力。
Herme, the talking robot developed at the Speech Communication Lab in Trinity College Dublin collected almost 500 signed consent forms from casual visitors to the TCD Science Gallery over a three-month period in 2012. Herme is a small LEGO device supporting a web cam and programmed to talk to any people who come within her range of view. She films the interaction and after chatting with her new 'friends' for about three minutes, tries to persuade them to sign an ethics committee consent form allowing the researchers in her lab to analyse the audio-video recordings of the conversation in order to better understand how people might interact with machines in the future. This demonstration showcases work being carried out at the Speech Communication Lab under SFI funding on the FASTNET project, with a Focus on Actions in Social Talk, linking computer speech synthesis with different forms of speech and image processing to make inferences about the success of an ongoing conversation. The corpus of recordings of conversations collected at the Science Gallery provides essential insights into how human social interaction follows clearly defined patterns that can be taken advantage of in advanced speech technology. The demonstration presented at ACII 2013 will allow participants to engage with Herme and judge her competence for themselves.