Yet we believe that there is an underlying commonality to all this diversity. Once it is understood and harnessed, it will not only open the door to even greater musical diversity but also augment creative expression. We call this underlying commonality generative potentials. We believe that generative potentials coexist in both music and in mind. This is a key focus of both our research and development.
We envision that our work will uncover at least three new paradigms that each will provide composers with new creative capabilities: ability to define generative building blocks and exploit their unique potentials, ability to directly express and control imagined (predicted) experiences, and deep artistic personalization not only to the individual artist, but to each new work.
1. Generative Building Blocks with unique potentials
The concept of motive as a musical building block with unique generative potentials that influence the working out of larger musical sections and structures is well understood in Western Classical Music. We believe this applies to all music. This is because our mind's innate generative mechanisms have already been demonstrated as universal in language, enabling the construction of novel messages that can be decoded and understood by their intended audience.
We hypothesize that the unique generative potentials of each musical building block is a result of a unique interrelationships between the specific musical features that comprise it. A change in one feature will change this potential unless it is possible to modify other features in a way that maintains it. Moreover, these potentials not only shape how larger sections and structures get worked out, but are also affected by the changing context over time and structure.
The implications on creative tools are twofold. First, if a composer edits a single feature within a building block, say moves a pitch up or down, then generative principles should dictate how other features should adapt in order to maintain the building block's generative potentials. Second, musical building blocks affect how larger phrases and sections get worked out. Therefore, if a composer adds, removes, or replaces a building block, then underlying generative principles should drive how the section they are embedded in is reworked.
2. Directly expressing and controlling musical experiences
One important goal of a musical composition is to generate in the listener a musical experience that unfolds as the music is being performed. In this sense, creation can be thought of as an attempt to encode in the music itself a musical experience that the composer is imagining (predicting). The fact that specific interrelationships exist between musical features and experienced feelings or emotions has been clearly stated in music theories as early as ancient Greece, in the Baroque doctrine of affections, through the modern Generative Theory of Tonal Music (GTTM). However, to effectively express, modify, combine, or morph between different emotive state requires an exceptionally high degree of musical competence and experience; there are no guiding principles that can help composers learn or ease execution.
If the musical experience, or emotive states, is directly encoded in the music itself, this means that generative potentials not only shape the working out of structures, but also shape the expression, control, and changes in the experience over time. We therefore hypothesize that musical generative building blocks contain not only musical features, but also emotive features, and that they are all interconnected through unique relationships. If our research could model this in a way that can also be embedded in software, this implies that composers could directly express and control emotive features and their evolution over time, analogous to current control over musical features such as pitch or rhythm.
3. Deep artistic personalization: the composer's thumbprint
Art is about manifesting the new and unique, often within well-defined constants or constraints. For example, it is hard not to tell apart Mozart from Salieri, as each composer's individual "thumbprint" is so unique. Yet they both share identical characteristics of the Classical Style, that, in some cases may even be interchangeable. These include melodic, rhythmic, harmonic, structural, and orchestral qualities. We hypothesize a dynamic model of music that has at least two interchangeable modules: one for genre, that is more stable and general, and another for each individual user, that is highly personalized and uniqe. Moreover, we foresee even deeper personalization as the user's module is further adapted to each unique creation. Thus, two users using the same tool with identical settings should produce outputs that differ according to these modules.