The ChoraleWind Dataset

logo_cw

Welcome to the accompanying website for the ChoraleWind dataset. On this website, we present some of the synthesized instrument tracks provided in the dataset. In the table below, pick a piece from the dataset. Using a web-based multitrack player, you have the possibility to create your own mixes by soloing and muting different tracks.

If you use ChoraleWind in your research, please cite the following paper:

  1. Axel Berndt, Aida Amiryan-Stein, Manuel Peters, Meinard Müller, and Stefan Balke
    ChoraleWind: An Expressive Wind-Quartet Dataset for End-to-End Rendering from the Neues Thüringer Choralbuch
    In Proceedings of the International Computer Music Conference (ICMC): 43–50, 2026.
    @inproceedings{BerndtAPMB26_ChoraleWind_ICMC,
    author    = {Axel Berndt and Aida Amiryan-Stein and Manuel Peters and Meinard M{\"u}ller and Stefan Balke},
    title     = {ChoraleWind: An Expressive Wind-Quartet Dataset for End-to-End Rendering from the Neues Th{\"u}ringer Choralbuch},
    booktitle = {Proceedings of the International Computer Music Conference ({ICMC})},
    address   = {Hamburg, Germany},
    year      = {2026},
    pages     = {43--50}
    }

Abstract

We introduce ChoraleWind, an expressive wind-quartet dataset and reproducible baseline for end-to-end rendering from the Neues Thüringer Choralbuch (NTCB). The dataset comprises 312 four-part chorales and covers the complete path from symbolic score encoding to performance-level rendition and synthesized audio. ChoraleWind includes a rule-based performance model that generates expressive timing, dynamics, and articulation, including metric and structural accents as well as phrase-end gestures from high-quality MEI encoding of the NTCB chorales, combined with controlled humanization, and physical-modeling wind-instrument synthesis that produces isolated voice stems and ensemble mixes. The dataset provides aligned symbolic representations, performance annotations, and multitrack audio, enabling systematic training and evaluation of score-to-audio wind-quartet rendering methods under fully controlled and reproducible conditions. Rather than aiming at state-of-the-art purely data-driven synthesis, ChoraleWind is designed as a transparent baseline and controlled testbed for studying expressive performance generation, timbre modeling, and reproducible evaluation of wind-quartet rendering systems.

Dataset

The table below lists 10 of the available works, the same ones used by ChoraleBricks. By clicking the title, you can create your own mix using the raw audio tracks provided in the downloads section below.

# ID Name Year Dur.
1 AN1 Anonymous: Aus meines Herzens Grunde 1598 00:46
2 BA1 Bach, J. S.: Ich steh' an Deiner Krippe hier 1736 00:37
3 CR1 Crüger, J.: Auf, auf, mein Herz, mit Freuden 1647 00:37
4 DR1 Drese, A.: Jesu, geh voran 1698 00:23
5 GE1 Gesius, B.: Befiehl Du Deine Wege 1603 00:50
6 GE2 Gesius, B.: Du Friedensfürst, Herr Jesu Christ 1601 00:40
7 JA1 Jan, M.: Du großer Schmerzensmann 1668 00:56
8 TE1 Telemann, G. P.: Der lieben Sonne Licht und Pracht 1730 00:46
9 VU1 Vulpuis, M.: Die helle Sonn leucht' jetzt herfür 1609 00:25
10 VU2 Vulpuis, M.: Christus, der ist mein Leben 1609 00:25

Note: The listed composers are the composers of the melody, but the four-part harmonization is always by Rudolf Mauersberger.

Downloads

Further Resources

License

CC BY 4.0 The data provided on this website is licensed under a Creative Commons Attribution 4.0 International License. Sheet music is published with the kind permission of the rights holder. If you publish results obtained using this data, please cite the paper mentioned above.

Acknowledgments

DFG Logo

This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Grant No. 500643750 (MU 2686/15-1) and under Grant No. 555525568 (MU 2686/18-1). The International Audio Laboratories Erlangen are a joint institution of the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) and Fraunhofer Institute for Integrated Circuits IIS.