The R3 Dataset

The R3 Dataset consists of ca. 31 hours of recordings from 62 different piano practice sessions by two professionally trained pianists. The recordings come from work on the Rach3 dataset (Cancino-Chac ́on and Pilkov, 2024) with additional processing and annotations added for the purposes of the related work on visual piano transcription (VPT) (Zivanovic et al., 2025).

Each session contains a topdown video with synchronized audio and captured MIDI recordings. Two different pianos are used throughout the practice sessions: a Yamaha GB1k Silent and an Essex EUP-116 equipped with a silent system (allowing for MIDI capturing). For audio we use AKG P170 condenser microphones and a Focusrite Scarlett 2i2 audio interface. The video is captured with GoPro cameras.

R3 includes two sub-splits which differ in content variety and size. The R3s split contains practice sessions from one pianist practicing Rachmaninoff’s Piano Concerto No. 3 Op. 30 (and hence the name of the dataset), while R3x contains sessions from both pianists practicing a variety of pieces from the Western Classical repertoire. Both splits have been processed exactly the same way and differ only in musical content.

The files published with this record contain the preprocessed dataset in hdf5 format for direct replication of our experiments as described in the related paper with the provided code and do not contain the full videos and/or audio. We also publish the full raw dataset at a separate repository The R3 Dataset (raw data) : Cancino-Chacón, C. E., Zivanovic, U., Pilkov, I., & Bishop, L. : Internet Archive for future work and for the benefit of the community.

Please check our paper for more comprehensive information.