Tuesday 23 October 2007

The Need for Music IR

Allister Heath and Jon Ashworth have written quite an interesting article on the massive changes going on in the music industry, and I wanted to pen some reflections on the implications of these changes with respect to music retrieval.



The evidence appears to show that more and more folk are buying their music online, and that CD sales are falling. Some artists are using a different model for generating revenue e.g. Prince released his recent album free with the Daily Mail, but did something like 21 dates at the O2 centre (that big tent thing) in Greenwich. Things could be going full circle here – musicians before the recording era earned their money from live performances. Things don’t look so good for the big labels (EMI, Sony BMG etc) let alone the music stores.



Hence Digital Rights Management (DRM) e.g iTunes. Music companies hope that this will stop their revenues from declining even further (evidence from the Heath/Ashworth article says that it would never replace CD sales). Personally I think they are going to be disappointed – for all sorts of reasons. If Prince and other artists want to give their music away freely, there is no need for DRM. Many DRM players are incompatible and users have to download specific software to use them. In some services (Napster), if you stop paying you are denied access to music you have already bought. In the long term I don’t think users are going to buy the DRM thing. Many voices e.g. Steve Jobs, have come out against DRM (at least for music). There is also very strong opposition to DRM from people like Richard Stallman and Ross Anderson. Frankly, I’m with them – DRM is mad!



It is my view that some stage in the future, we will have massive databases of music and will therefore have a big information retrieval problem (even if the music companies don’t release their back catalogue). How do we cope with this in terms of technical and user issues? Music IR has become something of a hot topic, and the ISMIR series of conferences and many of the technical challenges are being addressed including the process of matching and evaluation. Some of this research will prove very useful. However one important area is not being addressed sufficiently widely IMHO. Despite the excellent efforts of people like Sally Jo Cunningham, not enough research is being done into user needs for music. It’s all very well building sophisticated algorithms for matching, but how do you know they will actually service user needs. When attending an ISMIR conference, I remember being amazed by a statement by one computer scientist who claimed in all seriousness that ‘he knew’ what users wanted and didn’t need to talk to them. No, I didn’t believe him.



This is an important research direction for me. I have published in the area and have a student working on a project. Watch this space!



References
Heath, A. and Ashworth, J. (2007). “A brand new beat”, The Business, Available on line: http://www.thebusiness.co.uk/the-magazine/focus/269696/a-brand-new-beat.thtml [visited: 23rd October 2007]
Jobs, S. (2007). “Thoughts on music”, Available on line: http://www.apple.com/hotnews/thoughtsonmusic/ [visited: 23rd October 2007]

Web references
ISMIR: http://www.ismir.net/ [visited: 23rd October 2007]
MUSIC IR: http://www.music-ir.org/ [visited: 23rd October 2007]

No comments: