Letters

On Pleasing Every Palate

[Editor's note: This letter, written by the general chair of the Third International Conference on Music Information Retrieval (ISMIR 2002), responds to Maxwell Wells's review of that conference, which can be found on page 108 of this issue.]

Music information retrieval (MIR) has recently emerged as a common domain of inquiry in the collective consciousness of actors in diverse fields, including arts and humanities (music and other performing arts), behavioral and social sciences (cultural resources, broadcast media, economics, cognitive processes), engineering (acoustics, signal processing, ergonomics), law (communication law, international law, intellectual property), management and commerce, science and technology, and library sciences.

This is not to say that research, development, and uses of what is being now labelled as MIR had not existed previously. As Max Wells points out in his review, industry had been tackling MIR technology from its own point of view, with mixed returns (to say the least).

What is curiously absent from his historical review—other than his reference to the 1977 Tversky paper—is the fact that research and development had been going on in academia too for at least as long, with the publication of papers (as Downie's 1999 call and Foote's 1999 review paper show) in various conferences, periodicals and books (from psychology to computer music) and the invention of successful and usable systems (such as the Indiana University and IRCAM digital music libraries in the mid 1990s or Stanford University's Themefinder project later). Such topics as metadata have been in use in libraries much earlier than its relatively recent massive adoption by industry (actually, the first metadata records must have been written on Sumerian and Akkadian clay tablets; there are two primitive catalog tablets dating from around 2000 bce at the temple of Enlil at Nippur).

It is thus incorrect to state that—then as now—the publicized research "duplicated the work done by some of these companies." On the other hand, it is probably true that they had been led in parallel and with the added factor that, as Max Wells indicates, "few of the companies published, making it difficult to know what they did, or how they did it."

The establishment of ISMIR (see Byrd and Fingerhut 2002) aimed at remedying the lack of exchange which existed between the various earlier actors in MIR and creating a multidisciplinary forum bringing together the full spectrum of the MIR actors. This is a gradual process: the increase in attendance numbers is a good measure (from about 20 people at the early SIGIR '99 Exploratory Workshop on MIR, to the close to 200 attendees at ISMIR 2002) as well as an increase in the number of delegates from industry sectors (about 30% at ISMIR 2002): France Telecom, Nokia, NTT, Oracle, Philips, Sony, to name but a few, as well as many smaller commercial participants.

So the industry is here too—already, the telcos and audiovisual companies. Not all the relevant industry, as Max Wells points out: the recording industry isn't here yet. As a preliminary remark on the absence of a sector of the industry, one should say that it would be unmanageable for ISMIR to "grow too fast"—and the gradual reachout reflects a reasoned expansion, also reflected by the significant increase in the number of submissions and a widening of its areas of pursuit.

But then not all the other relevant actors are here either, such as law and intellectual property specialists, inasmuch as they apply specifically to MIR issues (the status and ownership of automatic audio summaries, of metadata, etc.). The ISMIR committee recognized this lack by inviting speakers to address those issues specifically in the invited guest speakers session on metadata: Chris Barlas of Rightscom, Ltd., spoke of the development of rights trading for music over networks and of the indecs analysis and MPEG 21 Rights Dictionary, following a talk by Leonardo Chiariglione of Telecom Italia and MPEG on the technologies of content representation, digital item declaration, interoperable intellectual property management and protection, and metadata.

As to the other, specific reasons for the absence of the recording industry, the factors which Wells mentions may soon lose their...