Conference paper Open Access
Zhengshan Shi; Craig Sapp; Kumaran Arul; Jerry McBride; Julius Smith
This paper describes the digitization process of a large collection of historical piano roll recordings held in the Stanford University Piano Roll Archive (SUPRA), which has resulted in an initial dataset of 478 performances of pianists from the early twentieth century transcribed to MIDI format. The process includes scanning paper rolls, digitizing the hole punches, and translating the pneumatic expression codings into MIDI format to create expressive performance files. We offer derivative files from each step of this process, including a high resolution image of the roll, a "raw" MIDI file of hole data, an "expressive" MIDI file that translates hole data into dynamics, and an audio file rendering of the expressive MIDI file on a digital piano sample. This provides digital access to the rolls for researchers in a flexible, searchable online database. We currently offer an initial dataset, "SUPRA-RW" from a selection of "red Welte"-type rolls in the SUPRA. This dataset provides roll scans and MIDI transcriptions of important historical piano performances, many being made available widely for the first time.