Fwd: [AUDITORY] test-set for singing voice segmentation

[Dan Ellis <dpwe@xxxxxxxxxxxxxxx>]

For our paper,

  A. Berenzweig, D. Ellis, S. Lawrence (2002)
  Using Voice Segments to Improve Artist Classification of Music
    Proc. AES-22 Intl. Conf. on Virt., Synth., and Ent. Audio. Espoo,
Finland, June 2002.
    http://www.ee.columbia.edu/~dpwe/pubs/aes02-aclass.pdf

we (well, Adam to be specific) hand-labeled a bunch of files for where
the vocals appeared.  These were the 15-sec samples that were randomly
recorded from the radio in 1996 by Eric Scheirer and Malcolm Slaney;
there were in total 64 that included vocals.  That seems like a
reasonable amount for evaluation, although more would be nice.

You can read about the data at
http://labrosa.ee.columbia.edu/sounds/musp/scheislan.html
If you want to download the data, I can send you the link.  I just try
to keep track of who downloads it.

We' ve always made these available (thanks to permission from Malcolm
and Eric) but I don't know if anyone else has actually evaluated
against them.  And I'm sorry if they've been a well-kept secret!

  DAn.


To:	AUDITORY@xxxxxxxxxxxxxxx,
Date:	Wed, Mar 5, 2008 at 8:13 AM
Subject:	[AUDITORY] test-set for singing voice segmentation
	
Dear all,

I wonder if there exists any public test set for testing singing voice
location algorithms,
or any available annotation of existing test set (such as RWC)
annotation into singing voice segments.

Thanks for your help

Geoffroy Peeters
Ircam - R&D
tel: +33/1/44.78.14.22
email: peeters@xxxxxxxx