overview . how to use it . architectural details . data . download
 
ConQuest - IJCAI'07

ConQuest-IJCAI'07 (Conference Questions IJCAI 2007) is a spoken dialog system that will provide schedule information during the IJCAI-2007 conference. The system knows about papers, sessions, workshops, and special events in the main conference. The system will also allow users to cast their votes for a People's Choice Best Paper Award. For more information about how to best interact with the system, click here.

Desktop machines running the ConQuest-IJCAI'07 system will be available at the conference venues from Saturday, January 6th for use by the conference participants.

Following the conference, we intend to make the data collected from the ConQuest: IJCAI system, freely available to the research community.

The ConQuest-IJCAI deployment is a joint effort between Carnegie Mellon University and International Institute of Information Technology, Hyderabad, India. A previous version of this system (see below) was initially developed by members of the CMU Speech Group and deployed during the Interspeech-2006 conference. More details about the system architecture are available here.


ConQuest - Interspeech'06

ConQuest (Conference Questions) is a spoken dialog system that provided schedule information during the Interspeech-2006 conference. The system know about papers, sessions, and special events in the main conference. The system allowed users to listen to announcements, and to cast their votes for a People's Choice Best Paper Award, which was presented during the closing ceremony.

We plan to make the entire system, as well as the data collected, freely available to the research community. Please check the download page.

The Conquest Interspeech-06 system was developed by members of the CMU Speech Group. The system is based on the RavenClaw/Olympus architecture. It uses a Sphinx-II speech recognition system together with a RavenClaw-based dialog manager, template-based natural language generation, and the Cepstral Swift synthesizer. More details about the system architecture are available here.

A similar system, DISCOH, has been developed by researchers at AT&T, ISCI and Edinburgh University for the SLT-2006 conference.