Sign in to view source links and access this dataset
Description
572 segments of pop music, approximately 3.5 hours of audio, with aligned multitrack MIDI transcriptions. The dataset was created by researchers including Nathan Pruyne and Chris Donahue, sourced from TheoryTab and the Lakh MIDI Dataset. The dataset page was last updated on July 10, 2026.
Use Cases
Train automatic music transcription models based on aligned multitrack MIDI data.
Analyze harmonic and instrumental structure in pop music based on the provided transcriptions.
Develop source separation algorithms using the multitrack MIDI alignments as ground truth.
Research music generation and arrangement using the segmented pop music data.
Strengths
Contains 572 music segments, providing a substantial collection for analysis.
Includes approximately 3.5 hours of audio with aligned transcriptions.
Features multitrack MIDI data, offering detailed instrumental separation.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Data may reflect genre and source bias inherent to the TheoryTab and Lakh MIDI sources.
Provenance
Source
TheoryTab and the Lakh MIDI Dataset.
Collection Method
Segments of pop music were sourced and aligned with multitrack MIDI transcriptions.
Freshness
Last updated 2026-07-10 19:11:28; freshness should be verified.
License is unknown; users must verify terms of use before downloading.