Computer Speech and Language Special Issue -- Call for papers

[Martin Cooke <m.cooke@xxxxxxxxxxxxxx>]

Computer Speech and Language Special Issue – Call for Papers

Speech Separation and Recognition Challenge

Guest Editors
Martin Cooke (University of Sheffield), m.cooke@xxxxxxxxxxxxxx
Te-Won Lee (University of California, San Diego), tewon@xxxxxxxx

A wide range of techniques exist for the separation – and subsequent 
recognition – of speech from other acoustic sources. Some approaches 
rely on signal processing and enhancement while others exploit 
statistical independence of acoustic sources. Some make strong use of 
speech knowledge to construct joint models of all sources present.

Until recently, there had been no extensive comparison of separation 
algorithms. However, a special session at Interspeech 2006 in Pittsburgh 
saw the results of the first Speech Separation Challenge – a large-scale 
attempt to evaluate different separation/recognition algorithms on the 
same task, which involved recognising simple sentences in the presence 
of other, similar, sentences.

This special issue of Computer Speech and Language solicits full papers 
describing systems tackling the Speech Separation Challenge, details of 
which are available at 
http://www.dcs.shef.ac.uk/~martin/SpeechSeparationChallenge.htm. 
Submissions are not limited to those who took part in the special 
session at Interspeech, but new entrants should contact the guest 
editors to clarify any details of the Challenge. Contributions 
describing both existing and novel methods for the recognition of 
separated speech are welcome so long as results are reported on the 
Challenge test data.

Important dates:
Submission deadline: June 30, 2007 (earliest submission: May 2007)
Completion of reviews: October 31, 2007
Revision deadline: January 15, 2008
Final decision: February 15, 2008

Electronic submission
Papers should be submitted through the Elsevier electronic submission 
system at http://ees.elsevier.com/csl and should follow the usual 
Computer Speech and Language guidelines (see links from the above URL). 
Papers should be identified as being intended for the special issue by 
choosing the article type as Speech Separation.