Content-Based Indexing of Symbolic Music Documents - Intelligent Music Information Systems: Tools and Methodologies - page 28

and relevance judgments. This is the goal of an
experiment that has been reported, where four
different approaches of melodic segmentation,
aimed at highlighting lexical units, have been
compared. Results showed that simple approaches
outperform more complex ones that exploit a
priori information either on music perception or
on music theory. Simple approaches are based on
the creation of a redundant index, where different
elements (i.e., a given note) belong to more than one
index term. From the experimental comparison it
may be inferred that redundancy is an important
aspect of music indexing. To this end, a final ex-
periment has been proposed, where a data fusion
approach has been exploited to mix together the
results of alternative indexing schemes. Results
showed that data fusion allows for an improve-
ment of the retrieval effectiveness.
From this discussion it can be concluded that
music indexing can inherit most of the advantages
of textual indexing, which is still a promising ap-
proach to music access, provided that the peculiari-
ties of the music language are taken into account.
Even if research in music retrieval should not be
limited to the extension of well-known techniques
to the music domain, indexing can be considered
as a core technique for building more complex
systems. Furthermore, it has to be noted that the
actual trends in the Music Information Retrieval
(MIR) research community encompass a number
of approaches that are beyond the pure retrieval
task. The user of a MIR system may have differ-
ent needs other than searching for a particular
song that she has in mind. The user may need an
information filtering system that allows the user
to listen only to the music documents that the
user (potentially) likes, or a browsing system for
managing a personal music collection. Also in
these cases, indexing can be used to increase the
efficiency of new approaches to music access.

references

Agosti, M., Bombi, F., Melucci, M. & Mian, G. A.
(2000). Towards a digital library for the Venetian
music of the eighteenth century. In J. Anderson,
M. Deegan, S. Ross & S. Harold (Eds.), DRH
98: Selected papers from digital resources for
the humanities (pp. 1-16). Office for Humanities
Communication.

Baeza-Yates, R. & Ribeiro-Neto, B. (1999).
Modern information retrieval . New York: ACM
Press.

Bainbridge, D., Nevill-Manning, C. G., Witten, I.
H., Smith, L. A. & Mc-Nab, R. J. (1999). Towards
a digital library of popular music. In Proceedings
of the ACM Conference on Digital Libraries (pp.
161-169).

Basaldella, D. & Orio, N. (2006). An application
of weighted transducers to music information
retrieval. In Proceedings of Electronic Imaging
(pp. 607306/1-607306/10).

Berenzweig, A., Logan, B., Ellis, D. P. W. &
Whitman, B. (2004). A large-scale evaluation of
acoustic and subjective music-similarity mea-
sures. Computer Music Journal , 28 (2), 63-76.

Birmingham, W. P., Dannenberg, R. B., Wake-
field, G. H., Bartsch, M., Bykowski, D., Mazzoni,
D., Meek, C., Mellody, M. & Rand, W. (2001).
MUSART: Music retrieval via aural queries. In
Proceedings of the International Conference on
Music Information Retrieval (pp. 73-82).

Blackburn, S. & DeRoure, D. (1998). A tool for
content-based navigation of music. In Proceed-
ings of the ACM Multimedia Conference (pp.
361-368).

Borko, H. & Bernier, C. L. (1978). Indexing con-
cepts and methods . New York: Academic Press.