panelist Archives - Gen9 Genetics
The problem we are trying to address here is most closely related to Tatsuya Kawahara's work on spoken language processing for conference and panel discussion audio files [4]. Kawahara's study provided an overview of various aspects of spoken language processing towards an intelligent archive of speech material. ...
... The compilation of these sentences could be a sufficient summary of the talk. In the generated file, the index phrases were hyperlinked to the segmented audio to facilitate navigation [4]. Xu et al. introduced the concept of audio keywords [5], which could be applied to the analysis of sports videos. ...
Video PowerSearcher: a text-based indexing e-learning system
Session document
Full text available
Jan 2010
Azzam SleitMoaath Hajaya
Farhan Obisat
In our rapidly evolving world, information and communication technologies have influenced the development of education. Both teaching and learning can benefit from new IT-based applications that save time and money and provide practical solutions for teachers and students. This article presents a prototype for a video search system called Video PowerSearcher. The system provides the ability to search within video conferences using text-based search queries. Video PowerSearcher separates audio from video content and performs preprocessing activities to identify video indexing functions that can be used at a later stage for search operations. Experiments were conducted on various video lectures using Video PowerSearcher to demonstrate desirable characteristics for e-learning systems, such as simplicity, speed, and accuracy.
View
... However, this article focuses on generating voice transcripts that are accurate and legible. In particular, we have been investigating the development of automatic transcription systems for conferences and meetings, which can be used for the generation of conference and meeting records [1]. Applications also include the Next Generation Transcription System for the Japanese Diet (Congress). ...
Intelligent transcription system based on spontaneous speech processing
Article
Full text available
Jan 2007
Tatsuya kawahara
With the improvement of speech recognition technology, the semi-automatic generation of conference and meeting document transcripts or records has become one of its promising applications. For this purpose, we must consider post-processing that includes cleaning up verbatim transcripts and segmenting into sentence and paragraph units. This article presents a novel statistical framework for an intelligent transcription system. It also reports on the recent progress of conference and meeting automatic speech recognition. Then, various approaches to the detection of speech units and the detection of disfluencies are described, as they are important in the post-processing of the transcripts generated by the speech recognizer.
View
... Automatic speech recognition (ASR) plays an important role in the production of these audio files because speech transcription is used to generate indexes or summaries that are essential parts of a file. We have been developing an automated filing system that includes ASR for discussion boards [1], [2]. A major problem in developing an ASR system dedicated to these types of speech is the difficulty of constructing a statistical language model that matches the target speech, because the amount of well-matched data is often limited. ...
Adaptation of PLSA-based language model of topics and speakers for automatic transcription of discussion boards
Article
Full text available
March 2005 SIICE T INF SYST
Yuya AKITA
Tatsuya KAWAHARA
Proper language modeling is one of the main problems for automatic transcription of spontaneous speech. We propose an adaptation method for statistical language models based on characteristics of both the subject and the speaker. This approach is applied for the automatic transcription of meetings and roundtables, in which