Reprinted from ICMC 2000 - 1 - Copyright  IBM Corporation 2000 A Framework for Representing and Manipulating Tonal Music Steven Abrams, Robert Fuhrer, Daniel V. Oppenheim, Don P. Pazel, James Wright {abrams, rfuhrer, music, pazel, jwright @watson.ibm.com } Computer Music Center IBM T. J. Watson Research Center P.O. Box 218, Yorktown Heights, NY 10598 Abstract This paper describes a novel set of tools, collectively known as “Smart Harmony”, that allows the composer to control and manipulate music at a high level, while constraining the music to conform to a designated functional tonal framework. The Smart Harmony system permits two essential types of operations. First, it allows musical material to be taken from one context and used in a new context – that is, with a new key and/or chord progression – and automatically transforms the material to sound appropriate in that context. Second, it enables musical material to be reshaped, pulling notes higher or lower in pitch, while constraining the material to sound appropriate within a given harmony. An important benefit of our mechanism derives from its relaxation of a strict interpretation of harmonic function, by which means we gain the flexibility to perform these manipulations. We describe data structures and algorithms to implement this mechanism, and offer motivation steeped in traditional functional harmony for its behavior. Lastly, we present musical examples. 1. INTRODUCTION The goal of our research in music composition systems is to better support the composer’s creative workflow. In order to more fully understand composers’ creative needs, we have engaged in extensive dialogs with many composers from different backgrounds. We have concluded from these dialogs that most systems fail to provide adequate mechanisms to: (1) capture a composer’s musical ideas quickly in a variety of forms, (2) manipulate them in musically sensible and interesting ways, and (3) organize those ideas in meaningful ways. More specifically, composers have expressed their need to quickly explore musical territory by reusing and reshaping musical materials within a tonal context. To that end, we have found a useful set of pitch-related musical manipulations that is not supported by current tools. This paper discusses one such family of musically interesting manipulations that are aimed at addressing this need, as embodied in our “Smart Harmony” system. The Smart Harmony conceptual framework was first designed by Oppenheim and prototyped in DMIX (Oppenheim: 1996), loosely based on a representation of Pachet (1993]) with the addition of functional elements of tonality. The current Smart Harmony implementation, described here, utilizes new data structures and algorithms that build on what we learned from the DMIX prototype. It is a central aspect of our Music Sketcher technology preview (Abrams et. al.:1999), which can be freely downloaded from http://www.research.ibm.com/music. Smart Harmony embodies a novel representation of functional harmony that models the functional role of pitches within the context of a tonal framework that consists of a harmonic progression and a tonal scale. The Smart Harmony system supports two basic manipulations: 1) Changing the chord progression associated with musical material, and automatically altering each pitch so that it retains aspects of its harmonic function in the new context. 2) Shifting musical pitches higher or lower, preserving aspects of their harmonic function. Because these transformations preserve certain aspects of the harmonic function of the individual notes, the overall result is that the altered pitches sound “appropriate”. These two operations offer a variety of useful and musically sensible possibilities, including: • changing the shape of a musical line of pitches by means of a time-varying curve, maintaining aspects of harmonic function • reusing musical material from one section or piece of music in another context, adapting it to conform to the new context Preservation of harmonic function is a fundamentally unique aspect of Smart Harmony. In order to support this, a pitch’s function in a particular context is explicitly indicated in its representation. Although our current representation does not capture harmonic function in its entirety, it enables the specific transformations mentioned above and provides room for future growth as well. In fact, there is a tension between preserving the harmonic function, register, and basic melodic shape of musical materials as they are transformed, and allowing the composer flexibility to change chord progressions, shift materials in register, and re-shape melodies. Smart