[Asis-l] MIREX 2008: Let the evaluations begin!

[J. Stephen Downie]

Hi gang:

Spring has sprung, the 2008 International Conference on Music
Information Retrieval (ISMIR 2008) papers are under review, and it is
May Day: Time to start defining the 2008 Music Information Retrieval
Evaluation eXchange (MIREX 2008) task set.

The MIREX 2008 wiki is available at:
http://music-ir.org/mirexwiki.

Please note that you must create a NEW login for the 2008 Wiki even if
you have a login that you previously used for editing the MIREX 2005 or
2006 or 2007 wikis.

Those new to MIREX might want to take a look at the MIREX 2007 results
summary (1 page, PDF) for inspiration:
http://music-ir.org/mirex/2007/abs/MIREX2007_overall_results.pdf

Also note that ISMIR 2008 (where MIREX will hold its plenary meeting) is
early this year (September 14th to September 18th, 2008;
http://ismir2008.ismir.net). This means we need to get all the runs done 
by end of August so participants can prepare their posters/abstracts 
prior to ISMIR. The traditional MIREX plenary meeting time has not be 
set yet but I will let you know as soon as I know.

As per usual, general, introductory MIREX announcements will be made to
the music-ir at ircam.fr and evalfest at mail.lis.uiuc.edu lists. As time
progresses, and conversations get more MIREX specific, MIREX traffic
will be posted primarily to the evalfest at mail.lis.uiuc.edu list. If you
are interested in MIREX, do subscribe to evalfest at mail.lis.uiuc.edu via:
https://mail.isrl.uiuc.edu/mailman/listinfo/evalfest.

Tasks are defined by the community so we need folks to step up and take
charge of putting specific tasks together. If we can get the data,
evaluation methods and THREE participants together, MIREX will do
everything it can to run a given task. Personally, I would love to have
some kind of chord detection task for this year: any volunteers? As of
today, MIREX will most likely rerun:

*Audio Cover Song Identification (great progress being made here!)
*Audio Mood Classification (an additional ground-truth set would be great)
*MultiF0 (both parts, I believe)

We are also working on framing a new classification task based upon a
collection of latin dance music (thanks to the team at U. of Kent, UK).

I have heard soft rumblings about some other tasks, so I now encourage
leaders (and interested parties) to move forward on:

*Score Following
*Optical Music Recognition

If you have any questions, please do not hesitate to contact me at:
jdownie at uiuc.edu.

On behalf of my labmates at IMIRSEL, I officially declare MIREX 2008 open!

Cheers, and be spiffy,
Stephen
-- 
**********************************************************
    "Research funding makes the world a better place"
**********************************************************
J. Stephen Downie, PhD
Associate Professor,
Graduate School of Library and Information Science; and,
Center Affiliate, National Center for Supercomputing Applications
University of Illinois at Urbana-Champaign
[Vox/Voicemail] (217) 649-3839
M2K Project Home: http://music-ir.org/evaluation/m2k