Future Work

Finally we will mention the main lines of research that remain open after finishing this Thesis. We will outline them in relation to the main contributions of the thesis: the CLAM framework, the DSPOOM metamodel, the OOCTM metamodel and the MetriX object-oriented music model.

In relation to the CLAM framework, in section A.1 we describe what are the main future lines of development. The main idea is that the framework's usability can still be improved in different ways and by doing so we believe that it will be accepted more naturally by more users. In order to do so we must address issues such as an easier deployment in any platform, a cleaner and clearer interface or the addition of more automatic flow control tools. All these issues are detailed in annex 6.5.

The Digital Signal Processing Object-Oriented Metamodel (DSPOOM) is complete and demonstrated through the CLAM framework. Nevertheless, we claimed that the metamodel is valid for any digital signal processing model and this still has to be tested thoroughly. Although some initial experiments confirm that the metamodel is useful in other domains such as image processing it should be tested in different situations and under different conditions. It is possible that the usage of the metamodel in different situations may end up reverting into the metamodel itself clarifying some details such as the convenience of using one graphical MoC or another (see section 4.3).

The Object-Oriented Content Transmission Metamodel (OOCTM) has been partially instantiated and demonstrated especially through content-based transformations (see 5.3.4). Nevertheless, the overall metamodel could be only proven through a rather limited application (see 5.4). These limitations are due to some restrictions especially on the analyzer process. This block addresses problems such as automatic instrument classification or musical transcription, which are still unsolved research topics. The full strength of the metamodel will only be visible when this technologies are finally available. Nevertheless we believe that its formulation will help in structuring related systems and already gives a conceptual framework for research in this area. In this sense it will also be important to exploit OOCTM's relation to other more well-stablished models/metamodels such as Structured Audio or S&W (see 5.3).

Finally we already commented that the MetriX music model presented in chapter 6 does not intend to be as general as the other metamodels previously presented. MetriX is used as a proof of concept and demonstration that the object-oriented paradigm and the DSPOOM metamodel in particular may be used to effectively model the symbolic music domain. Nevertheless, we believe that the resulting model could with minor enhancements become of general applicability. The major shortcomings of the models are that it lacks a way of describing performers and that analysis information is not fully integrated.

Concentrating on the former limitation we observe that any music performance is made of at least three main actors: the score, the instrument, and the performer. The performer reads and interprets the score and acts on the instrument accordingly. The definition of Score in our OO music model is a hybrid between the traditional score written by a composer and the interpretation given by the performer. We would like to have a more clear distinction between the three levels and would therefore be interested in defining a third ``document'' specifying how a performer behaves. This file would contain a list of constraints to be applied to the way score events would be sent to the instrument.

As for the second limitation, it would be interesting to better integrate MetriX with our OOCTM. In order to do so new ways to include analysis information into both the Score and the Instrument definition need to be devised. By doing so we would be accomplishing a more compact conceptual integration of the three main axis in this thesis: CLAM/DSPOOM, OOCTM and MetriX.

It is clear that any piece of work leaves many open issues and lines of future research. It is even more so in a Thesis such as this where models of very general applicability are presented. Nevertheless it is important to ask oneself whether the ground that has been built is solid enough so as to continue constructing from it. In this sense we believe that we have accomplished our goals and we hope to have contributed to the advance of not only present but also future research in our field.