Pen-Based Interaction for Intuitive Music Composition and Editing - Intelligent Music Information Systems: Tools and Methodologies - page 271

Figure 6. Drawings of a filled-note head (on the
left) and a sharp (on the right)

a shape recognition ( SR ) block and a document
context creation ( DCC ) block, in the following
order:

SymbolName (Parameter 1 , ... , Parameter n)
Document Context Verification block (DCV).
Shape Context Verification block (SCV).
Shape Recognition block (SR).
Document Context Creation block (DCC).

The formalism we propose takes all these
concepts into account. We present its syntax and
explain it on two classical musical score notation
examples, which are the filled-note head and the
sharp.
The description of the interpretation of a struc-
tured document with our formalism is composed
of interpretation rules which define the generating
of the element which name they bear. Several
rules can have the same name, which makes it
possible to model different ways to compose the
same element. A rule takes a set of elements as
parameters, enounced in its heading, and returns
a new one that can replace them; the parameters
are the components , or sub-elements , of the new
element. A parameter can be either a stroke or an
already interpreted element, which makes possible
a hierarchical shape description. For instance, as
presented in Figure 6, a filled-note head is typi-
cally drawn with one stroke, whereas a sharp is
typically drawn with two horizontal segments
and two vertical segments. The corresponding
rule headings can then be:

DCV and DCC blocks enable a global vision of
the document in order to define in which document
structural contexts an element must be located.
A DCV block specifies the document structural
contexts in which the symbol created by the rule
has to be identified, whereas a DCC block indicates
the contexts that are generated due to the creation
of this element. The SCV and SR blocks enable,
given a document context, a local vision of the
element to recognize; it distributes the recogni-
tion process among local constraints, formalized
in the SCV block, and recognizers, formalized in
the SR block; it drives the interpretation process
by calling dedicated recognizers depending on
the context of elements.
The proposed formalism is based on the defini-
tion of structural contexts which model, on the one
hand, specific locations in the document, and, on
the other hand, which elements can or must exist
at these positioning. The syntax of a structural
context is as follows:

R[position,part]A.

FilledNoteHead (Stroke s) …

This means that the involved structural con-
text is located at the relative positioning position
(e.g., in, on the left, above, etc . ) of a reference
R . In order to satisfy this context, an element A
must have its part part (e.g., one point, all the
points, the first point, the highest point, etc . ) in
this positioning.

Sharp (HorizontalSeg hs1 , HorizontalSeg hs2,
VecticalSeg vs1 , VecticalSeg vs2) …

The structure of a rule is composed of four
blocks: a document context veriication ( DCV )
block, a shape context veriication ( SCV ) block,