Automated Notation of Piano Recordings for Historic Performance Practice Study

作者：Trevino Jeffrey^*; Sapp Craig

来源：Journal on Computing and Cultural Heritage, 2014, 7(3): 17.

DOI：10.1145/2597179

• Music
• Recording
• Performance
• Notation
• Visualization
• Automation
• visualization
• mapping
• pedagogy
• historic performance practice

We describe a system that automatically notates a comparative visualization of multiple recorded performances of the same musical work. Written musical scores have transmitted basic performance information to musicians over the ages; however, these scores only provide skeletal instructions that must be fleshed out in performance, as musical notation describes phrasing, articulation, dynamics, accentuation, and other ornamentations in generalized and ambiguous forms. Consequently, musical performances derived from the same notation can vary widely from each other in the same manner that a written text may be spoken with intense emotion or in flat monotone. Prior to the advent of recording technology, musical performances were ephemeral, only occurring once, never to be heard again in exactly the same rendition. As a result, musical interpretations were informed only by live listening. Now, with more than a century of recorded performance practice, musicians can delve deeper into the history of their aural art to gain inspiration and insight from sources that would otherwise have been inaccessible. Performers have become interested in giving performances inspired by recordings of the past, which often obey a musical common sense alien to the standards of modern practice, and it is useful for historically informed performers to describe, analyze, emulate, and internalize the performance styles of the past through the detailed study of recordings. Although much can be learned by listening, a visual interface may reveal potentially inaudible details of a recording. Because performers interact daily with traditional musical notation-a sophisticated, if ambiguous, multidimensional visualization of musical information-one approach to the design of such an interface leverages performers' existing knowledge by reducing the gap between data visualization and traditional musical notation as much as possible. Using Abjad, a Python-based tool for musical composition, the symbols of conventional staff notation are augmented to illustrate the intensity and temporal proximity of performed musical events graphically, thus facilitating the comparison of individual performances and the study of changes in performance aesthetics over time.

• 出版日期2014-8
• 单位MIT