Short description:

This showcase demonstrates searching/browsing in Audio and Video-content of NOS 8 o'clock journal using multimedia retrieval technology. Two options can be used:

The classical way: in journal of the last 4 weeks, using Teletext subtitles, or using the audio (which is transferred to text) using a Dutch speech-recognizer. All work is done automatically: the video-file is cut automatically in segments based on changing the item (teletext) or changing the speaker (speech-recognition). Using Google Maps you can browse through the 8 o'clock journal of the latest 4 weeks. Places are found in the teletext or subtitles information. NB. This demonstrations is created using work from the students: Joris van Balen, Tim Klifman,Peter Vel and Tijn van Hofwegen

Keywords:

multimedia retrieval broadcast news speech recognition audio mining spoken document retrieval

Research Themes:

• Multimedia Retrieval
• Speech and Language Technology

Description

The teletext information dated september 2 2011 until september 24 2011 is not present

Classic search	Browse with Google Maps
Input the subject you want to search for: Search-item(s) Information from: Teletext (888) ASR Shout(*) ASR Abbot(*) Channel: journaaltv-oost (*) Shout and Abbot are Speech-recognizers	Teletext (888)Speech-recognizer
Classic search	Browse with Google Maps

HMI-members working on this showcase are:

• Arjan van Hessen
• Franciska de Jong
• Hendri Hondorp
• Roeland Ordelman

• Marijn Huijbregts

Projects involved with this showcase:

• MultimediaN [MultimediaN/N5]
• Waterland [Workflow with semi-Automatic meTadata Extraction for fully digital media pRoduction in the NetherLANDs]