文章基本信息

标题：Hmi Modelling for Multimodal Lithuanian Applications
本地全文：下载
作者：Rytis Maskeliunas ; Kastytis Ratkevicius
期刊名称：Public Policy And Administration
印刷版ISSN：2029-2872
出版年度：2012
卷号：41
期号：2
页码：136-142
DOI：10.5755/j01.itc.41.2.909
语种：English
出版社：Kaunas University of Technology
摘要：Spoken dialogue based human-machine interfaces (HMI) are becoming more and more widely integrated in computer applications. Speech allows doing some task easier and faster. The combination with a more traditional means of input and out – i.e. the multimodality factor becomes more and more important allowing wider accessibility. It is important to model and design spoken language dialog trees to mimic the natural conversations in the human-computer interactions, especially in information retrieval systems and applications. The paper presents three algorithms of HMI dialogs and the results of their experimental evaluation. The results showed that it is possible to achieve about 97% recognition accuracy in simple phrase based dialog conversations and about 93% in a very naturally sounding keyword spotting based dialogs.
关键词：speech recognition;voice dialog modeling;human-machine interfaces