Editor's Message to Special Issue on Spoken Document Processing

Abstract

Recently, speech recognition technology became popular such as in applications of voice search, speech translation and TV closed caption. Until two decades, researchers and engineers were concerned with how to spread speech recognition technology. When we think of those days, we feel like we are living in a completely different age. Needless to say, it was caused not only by the progress of speech recognition technology, but also endeavor of company, because it is necessary further devise the technology for transferring to a general market. When the technology matched with citizen’s demand, the market will expand rapidly. We can say that the speech recognition technology will get better with age. People say the technical term of “speech recognition” in daily conversation. This is influenced by the commercial of speech recognition applications such as other goods. From a view point of this phenomenon, it looks like the research of spoken language processing is nearly complete, but it is not true. The biggest user demand for speech applications is to improve the speech recognition performance and to make it robust for various environments. Furthermore there still remains a problem how to utilize big data of spoken documents. Significant Interest Group of Spoken Language Information Processing (SIG-SLP) of Information Processing Society of Japan founded “Working Group of Spoken Document Processing” at 1996 and finished in 2012. During that period, the group set the task of spoken retrieval, recorded the spoken documents and published the transcription. At the same time, we held “Workshop of Spoken Document Processing” every year together with Toyohashi University of Technology. In accordance with the progress of speech recognition technology and the increase of spoken documents, the researches related to spoken documents have become more active. Therefore we thought the special issue was valuable for developing such fields and published it firstly in 2009. This is the second special issue. Topics of this issue are related to spoken document/corpus processing of lecture, meeting, broadcasting, spoken dialog, speech translation and so on. We informed the special issue and CFP at the homepage of IPSJ, meeting of SLP and mailing-list. We had 21 submitted papers (29 papers for the first issue) and accepted 11 papers including 3 English papers (13 papers for the first issue). Out of these accepted papers, 5 papers were related to spoken retrieval, because of the influence of the Working Group.