| 題目：Spoken Content Retrieval - Beyond Cascading Speech Recognition with Text Retrieval 主講人：李宏毅 教授(國立台灣大學 電機系) 時間：103年10月1日(星期三13:30-15:00) 地點：三峽校區法學院大樓2F01教室 Abstract: Today the multimedia content over the Internet has become a part of human life, and the spoken part in multimedia content very often describes the core concept of the multimedia. Although the public search engines are very successful in searching for web pages based on the text, searching directly over the spoken content is still difficult. However, retrieval directly over the spoken content not only eliminates the needs to provide text information for the multimedia for indexing purposes, but can locate the exact utterance for the desired information.Over the last decade, spoken content retrieval has achieved significant advances by primarily cascading speech recognition techniques with text information retrieval techniques. This approach works well when the recognition accuracy is high enough, but becomes less adequate for more challenging real-world tasks such as retrieving course lectures or telephone conversations with relatively low recognition accuracy, because the retrieval performance is highly dependent on the speech recognition accuracy.Is it possible to have better retrieval results under relatively low speech recognition accuracy? In this talk, I am going to introduce some innovative approaches of spoken content retrieval beyond the mentioned cascading framework with retrieval performance shown to be less constrained by speech recognition accuracy. For example, the retrieval system can exploit the information in speech signals not present in the output of the standard speech recognition modules, speech recognition system can be redesigned to have better retrieval performance or even be eliminated from the spoken content retrieval process, the users can navigate across the desired spoken content with dialogues, and the search results can be visualized in an intuitive interface, etc.. Short Bio: Hung-yi Lee received the M.S. and Ph.D. degrees from National Taiwan University (NTU), Taipei, Taiwan, in 2010 and 2012, respectively. From September 2012 to August 2013, he was a postdoctoral fellow in Research Center for Information Technology Innovation, Academia Sinica. From September 2013 to July 2014, he was a visiting scientist at the Spoken Language Systems Group of MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). He is currently an assistant professor of the Department of Electrical Engineering of National Taiwan University, with a joint appointment at the Department of Computer Science & Information Engineering of the university. His research focuses on spoken language understanding and speech recognition. |
