Abstract
In the digital age, reproducing piano rolls are increasingly hard to hear; the original playback technologies are rare, and many have deteriorated since their construction. Thankfully, a vast archive of digitisations has enabled performances, captured on such rolls, to become increasingly available for use within musical research. These digitisations preserve historic performance practices that have, in many cases, since disappeared from use and, as a consequence, they have an extremely valuable role to play in numerous research fields. The nature of this role may, however, be questioned; digitisations are often taken to be primary sources of evidence, seemingly offering direct and immediate access to once-upon-a-time performances. The processes involved in their production, however, suggest that they may be more appropriately understood as secondary sources. To demonstrate this point, this article offers a case study through which the production of digitisations is scrutinised; a range of visualisation tools are used to examine nine digitisations of a single piano roll and, although one may expect uniformity among the digitisations, visualisations reveal significant and substantial differences. Some of these differences may be attributed to the piano roll technologies, particularly in terms of the voicing and balance of the piano. Others are a consequence of the digitisation process itself; specific recording techniques, room acoustics, and microphone selection are but some of the many variables that determine the nature of the digital result. As the article develops, it becomes increasingly clear that digitisation profoundly influences what we hear. It is paramount that we understand the variables involved in the production of digitisations, since their capacity to take us back to the past ensures that they remain invaluable sources of evidence long into the future.