Odyssey 2008: Presentations
The Odyssey 2008 Presentations are now available. Any authors who still wish to make their presentations available should send them to odyssey08review@gmail.com.
Odyssey 2008: Scientific Program (updated on 2008/01/08)
The updated scientific program is available here:
Odyssey 2008 Scientific Program
Odyssey 2008: Presentations
The Odyssey 2008 presentations are now available. Any authors that still wish to make their presentations available should send them to odyssey08review@gmail.com.
Odyssey 2008: Registration open
Registration for Odyssey 2008 is now open. Please refer to the Conference Registration and Venue pages for information on registering for the conference and for accommodation at the venue.
The workshop venue is a hotel & conference center on a wine estate called Spier (www.spier.co.za) about 10 km from Stellenbosch, about 40 km from Cape Town, and about 30 km from Cape Town International Airport, South Africa.
We have made a block booking at Spier at much reduced room rates -- details to follow soon.
The dates of the workshop will be 21-24 January 2008. We have shortened the conference from 5 to 4 days. January is high summer in the Southern Hemisphere.
Odyssey 2008: Review Process Completed
The review process is now complete. Authors may retrieve paper acceptance decisions and reviews using the paper status link. (You will need your paper ID which was mailed out on paper submission.) Please let us know at odyssey08review@gmail.com if you have any problems with retrieving your paper status.
You may also refer to the list of papers accepted for Odyssey'08.
Authors of papers which have been accepted have until 30 October 2007 to modify papers according to the recommendations of the reviewers. Please also keep in mind that papers may be up to 8 pages long. Final papers may be uploaded by using the Paper submission link.
Odyssey 2008: Call for Papers
About Odyssey - The Speaker and Language Recognition Workshop Series
Research and development on speaker recognition methods and techniques has been undertaken for well over four decades, and it continues to be an active area. Approaches have spanned from human aural and spectrogram comparisons, to simple template matching, to dynamic time-warping approaches, to more modern statistical pattern recognition approaches, such as neural networks and Hidden Markov Models (HMMs).
It is interesting to note that, although striving to extract and recognize different information from the speech signal, many of the same features and techniques successfully applied to speech recognition have also been used for speaker recognition. Over this same time, research and development corpora have evolved from small, private corpora (5-10 speaker) under laboratory clean, controlled conditions (single session, read speech) to large, publicly available corpora (500