Information Technology Reference
In-Depth Information
tries to emerge a classification from the
database, by clustering songs according to
a given measure of similarity. While the
prescriptive approach adopts the framework
of supervised learning, this second point-
of-view is unsupervised. Another important
difference is that in the first approach, genre
classifications are considered as natural
and objective, whereas in this approach it
is similarity relations which are considered
as objective.
music (aural, visual, logical, structural descrip-
tions). But this key advantage has an interesting
consequence: the possibility to create a strongly
interconnected and synchronized environment
to enjoy music. The purpose of our MXDemo is
illustrating the full potentialities of an integrated
approach to music description.
This goal can be achieved thanks to three
cooperating elements:
•
A comprehensive XML-based format to
encode music in all its aspects.
•
A software environment aimed to the in-
tegrated representation. The software ap-
plication will provide a graphic interface
to read, watch, and listen to music, keeping
the different levels synchronized.
•
An automatic system to synchronize music
score and the related audio signal.
audIo-score automatIc
sYnchronIzatIon
Music language is made up of many different and
complementary aspects. Music is the composition
itself, as well as the sound a listener hears, and
is the score that a performer reads as well as the
execution provided by a computer system.
The encoding formats are commonly accepted
and employed and often characterized by a partial
perspective of the whole matter: they describe
data or metadata for score, audio tracks, computer
performances of music pieces, but they seldom
encode all these aspects together. Nowadays,
we have at our disposal many encoding formats
aimed at a precise characterization of only one
(or few) music aspect(s). For example, MP3,
AAC, and PCM formats provide ways to encode
audio recordings; MIDI represents—among other
things—a well known standard for computer—
driven performance; TIFF and JPEG files can
result from a scanning process of scores; NIFF,
Enigma, Finale formats are aimed at score typing
and publishing.
The first problem to face is finding a compre-
hensive way to encode all these different aspects
in a common framework, without repudiating the
accepted formats. An important advantage of such
effort is keeping together all the information re-
lated to a single music piece, in order to appreciate
the richness of heterogeneous representations of
About point number one, some examples of
XML-based formats used to encode music are
presented in Haus and Longari (2002) and Roland
(2002). They are not discussed in this section.
About the second one, Kurth, Muller, Rib-
brock, Roder, Damm, and Fremerey (2004)
propose a generic service for realtime access to
context-based music information such as lyrics or
score data. In our Web-based client-server sce-
nario, a client application plays back a particular
(waveform) audio recording. During playback, the
client connects to a server which in turn identifies
the particular piece of audio as well as the current
playback position. Subsequently, the server deliv-
ers local, that is, position specific, context-based
information on the audio piece to the client. The
client then synchronously displays the received
information during acoustic playback. Kurth et
al. (2004) demonstrates how such a service can be
established using recent MIR (Music Information
Retrieval) techniques such as audio identification
and synchronization.
Baraté (2005) proposes the MXDemo, a
stand-alone software which illustrates the full
Search WWH ::
Custom Search