Conclusions and Future Work

In this Thesis we have presented a metamodel for digital signal processing named DSPOOM.

This metamodel uses the Object-Oriented paradigm and exploits the relation between this paradigm and the Graphical Models of Computation used mainly in system engineering. Because these and other concepts used in the Thesis are borrowed from different areas of knowledge it is first necessary to establish a coherent corpus and conceptual framework. This is introduced in chapter 1. We believe that this chapter cannot be only understood as a conceptual introduction, though. The way that the different concepts are presented and related in between them constitutes a set of hypotheses in itself, hypotheses that are then elaborated and demonstrated in the following chapters. The main hypothesis can be summarized in: ``Systems are object-oriented''; ``Signal processing systems can be modeled through object-oriented and graphical models of computation''; and ``Frameworks generate metamodels''.

A metamodel cannot be defended in any other way than through the evaluation of the models it generates and their usefulness for modeling systems in a particular domain. In this Thesis this is accomplished through the CLAM framework. CLAM, which is a framework for music and audio processing, is as a matter of fact the origin of the DSPOOM metamodel that was obtained as an abstraction of the framework and the different models included in it. In chapter 3 we have presented the main features of the CLAM framework and we have compared them to some included in similar frameworks. For doing so we have had to first present a thorough review of similar environments in chapter 2. As already pointed out in the corresponding chapter we believe that CLAM represents a different approach based especially on the use of good software engineering techniques.

The Digital Signal Processing Object-Oriented Metamodel (DSPOOM) is an abstraction of different conceptual conclusions we reached during the framework development process. Nevertheless it shares many concepts and ideas with the frameworks reviewed in this thesis. It presents a clear way to model a signal processing system while it enhances the separation between data and process, and idea that is intuitive to signal processing engineers. It also offers and explicit graphical Model of Computation that helps in modeling complex systems and understanding them.

The Object-Oriented Content Transmission Metamodel is a particular instance of DSPOOM that can be used to model the so-called ``Content-based applications''. The OOCTM is also demonstrated through some particular applications that partially instantiate the metamodel. The applicability of the whole metamodel is still not guaranteed as it depends on some ongoing research that is supposed to develop the necessary technology for some of the building blocks. Nevertheless we believe that the conceptual framework is useful in itself and will help in understanding also the practical applicability of the metamodel.

Finally the object-oriented music model presented through the MetriX language is a particular model that can be understood as a proof of concept that object-orientation in general and DSPOOM in particular can be used to effectively model more abstract domains such as music. Although the scope of this music model is not as broad as to give response to any musical application and its implementation presents some limitations, we believe it is generic enough so as to become useful in most situations with a few minor additions (see again section 6.5 on Future Work)

After this concluding conceptual summary we will finish this section by completing the glossary given at the Introduction and by outlining possible lines of future work.