PIG Dataset      
PIG Dataset is publicly available data for academic research on piano fingering. It is a collection of classical piano pieces with fingering annotations for all musical notes. Upon the use of PIG Dataset, we kindly ask every user to explicitly cite/mention the following paper. Thank you for your understanding and we hope that you find the dataset useful.
Eita Nakamura, Yasuyuki Saito, Kazuyoshi Yoshii
Statistical Learning and Estimation of Piano Fingering
Information Sciences, Vol. 517, pp. 68-85, 2020. [arXiv:1904.10237]
Accompanying webpage


Only nonprofit, academic use is allowed.
Please go to the DOWNLOAD PAGE.


(22/5/12, version 1.12) Minor corrections in fingering data for 029-5, 032-2, and 032-3. Many thanks to Mr David Randolph for finding the errors.

Data content

Data format

The dataset contains musical scores in PDF and one or more fingering files in the text format for each musical score. A fingering file is a list of musical notes and each note has the information of onset time, offset time, pitch, and finger number. For a more detailed description of the data, please see README.pdf in the dataset.

List of musical pieces

There are currently 150 pieces by various composers and each piece is given one or more fingerings. In total there are 309 distinct fingerings provided by various pianists. There are special subsets that are intended for studying individual differences of fingerings. Each piece in these subsets is given fingerings by at least four different pianists. These consist of 10 pieces by J. S. Bach (Bach set), 10 pieces by W. A. Mozart (Mozart set), and 10 pieces by F. Chopin (Chopin set). The set of the rest of the data is called the miscellaneous set.
Complete list of musical pieces (PDF)


Picture This visualizer can be used for observing and comparing the fingering files in the dataset. It works on browsers such as Google Chrome, Safari, Firefox, Microsoft Edge, etc.


For any inquiries or feedback of PIG Dataset, please contact one or both of the following persons.


We would like to thank Dr Shinichi Furuya and Dr Kazuyoshi Yoshii for their cooperation in data preparation. We are also grateful to many other cooperators who contributed to providing fingerings and setting up data.

This project is partially supported by JSPS KAKENHI Nos. 15K16054, 16H02917, 16K00501, and 16J05486.