Music Representation of Score, Sound, MIDI, Structure and Metadata All Integrated in a Single Multilayer Environment Based on XML - Intelligent Music Information Systems: Tools and Methodologies - page 311

• Logic: The symbols that compose the score

as intended by its author(s)

layer; on the contrary, MP3 files and other audio
descriptions are not translated into XML format,
rather they are linked and mapped inside the cor-
responding MX layer, namely Audio layer.
Needless to say, not all layers must or can be
present for a given music piece. Of course, the
higher their number, the richer the piece's descrip-
tion is. Besides, each layer could contain one or
more object instances: for example, Audio layer
could link 3 audio tracks and Structural layer
could provide two different analyses for the same
piece. The concept of multilayer description (i.e.,
potentially many different types of descriptions,
all correlated and synchronized) together with the
concept of multiinstance support (i.e., potentially
many different media objects for each layer) pro-
vide a very rich and flexible means of encoding
music in all its aspects. Section 5 will illustrate
possible applications of these characteristics.
MX layers will be soon analyzed in greater
detail and made clearer by some example, but
first we have to introduce the main concept MX
is based on: the idea of spine.

• Structural: Music objects and their relation-

ship

• Notational: Graphical representations of

the score

• Performance: Computer-based descriptions

and executions of music

• Audio: Digital or digitized recordings of

the piece

Figure 2 shows the relationship between the
group constituted by General, Structural, and
Logic layers and the one encompassing Nota-
tional (e.g., GIF, JPEG, TIFF for a score), Audio
(e.g., AAC, MP3, WAV), and Performance (e.g.,
Csound, MIDI, MPEG) layers. It should be evident
that our approach consists in keeping intrinsic mu-
sic descriptions inside the MX file (the upper block
in Figure 2), whereas media objects are external
(the lower block in Figure 2). Let us cite some
clarifying examples. The symbols that belong to
the score, such as chords and notes, are directly
described in XML format, in the so-called Logic

spine

Figure 2. Relationships among MX layers and
external media objects

According to our comprehensive approach to
music description, the MX standard provides dif-
ferent layers to represent information. Of course,
these levels cannot be independent: in order to
achieve richness in information description, each
music event can have one or more descriptions
in different layers. Thus, we should be able to
create a common data structure in order to link
the different representations of each music event
within the MX file. For instance, let us refer to a
hypothetical event named e12, corresponding to
a B quaver somehow described in Logical layer:
we should be able to refer to its representation also
in audio, video, score parts of the MX file.
Accordingly, we have introduced the concept
of spine, which is an overall structure that relates
time and space information. Spine contains no