Schubert Winterreise Dataset

The Schubert Winterreise Dataset (SWD) is a multimodal dataset comprising various representations and annotations of Franz Schubert's 24-song cycle Winterreise. The primary material (raw data) consists of textual representations of the songs' lyrics, music scores in image, symbolic, and MIDI format, as well as nine audio recordings of performances (only two included due to copyright issues). The secondary material (annotations) comprises information of musical measure positions in sheet music images and audio recordings as well as analyses of chords, local keys, global keys, and structural parts.

UPDATE: From version 2.0 onwards, the SWD contains additional note annotations comprising pitch, pitch class (chroma), and instrument labels for each note event (generated semi-automatically).

The SWD is organized as follows:

01_RawData

lyrics_txt

score-IMSLP_png

score-IMSLP_pdf-complete.pdf

score_sibelius

score_pdf

score_musicxml

score_midi

audio_wav

02_Annotations

ann_score-IMSLP_measure

ann_score_chord

ann_score_localkey-ann1

ann_score_localkey-ann2

ann_score_localkey-ann3

ann_score_globalkey.csv

ann_score_structure

ann_audio_measure

ann_audio_chord

ann_audio_localkey-ann1

ann_audio_localkey-ann2

ann_audio_localkey-ann3

ann_audio_globalkey.csv

ann_audio_structure

ann_audio_note

03_ExtraMaterial

original data

scripts for cutting

licenses of the published performances

Further information can be found in the README file and the accompanying journal paper.

License:

The dataset is published under a Creative Commons Attribution 3.0 Unported license (https://creativecommons.org/licenses/by/3.0/legalcode). We acknowledge the publication of the performance by baritone Randall Scarlata with pianist Jeremy Denk (denoted as SC06), published under the same license by the Isabella Stewart Gardner Museum (Boston). Please note that the audio files of this performance must not be re-distributed in modified form. All other sources for the dataset are in the Public Domain.