Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PDMX: A Large-Scale Public Domain MusicXML Dataset for Symbolic Music Processing

Phillip Long, Zachary Novack, Taylor Berg-Kirkpatrick, Julian McAuley

TL;DR

PDMX addresses the lack of large-scale, public-domain symbolic music data by creating a CC-0 MusicXML dataset with over 250k scores from MuseScore. The authors introduce MusicRender, an extension of MusPy, to faithfully parse and render MusicXML cues, and they systematically analyze data quality via rating metadata and deduplication. Through experiments with a REMI+-based transformer on various data subsets, they show that rating-driven filtering and deduplication improve harmonic and rhythmic modeling, with fine-tuning on high-quality data further enhancing performance and perceptual quality. The work provides a scalable, license-safe resource with rich metadata for both generative modeling and discriminative MIR tasks, enabling safer pretraining and downstream applications in symbolic music processing.

Abstract

The recent explosion of generative AI-Music systems has raised numerous concerns over data copyright, licensing music from musicians, and the conflict between open-source AI and large prestige companies. Such issues highlight the need for publicly available, copyright-free musical data, in which there is a large shortage, particularly for symbolic music data. To alleviate this issue, we present PDMX: a large-scale open-source dataset of over 250K public domain MusicXML scores collected from the score-sharing forum MuseScore, making it the largest available copyright-free symbolic music dataset to our knowledge. PDMX additionally includes a wealth of both tag and user interaction metadata, allowing us to efficiently analyze the dataset and filter for high quality user-generated scores. Given the additional metadata afforded by our data collection process, we conduct multitrack music generation experiments evaluating how different representative subsets of PDMX lead to different behaviors in downstream models, and how user-rating statistics can be used as an effective measure of data quality. Examples can be found at https://pnlong.github.io/PDMX.demo/.

PDMX: A Large-Scale Public Domain MusicXML Dataset for Symbolic Music Processing

TL;DR

PDMX addresses the lack of large-scale, public-domain symbolic music data by creating a CC-0 MusicXML dataset with over 250k scores from MuseScore. The authors introduce MusicRender, an extension of MusPy, to faithfully parse and render MusicXML cues, and they systematically analyze data quality via rating metadata and deduplication. Through experiments with a REMI+-based transformer on various data subsets, they show that rating-driven filtering and deduplication improve harmonic and rhythmic modeling, with fine-tuning on high-quality data further enhancing performance and perceptual quality. The work provides a scalable, license-safe resource with rich metadata for both generative modeling and discriminative MIR tasks, enabling safer pretraining and downstream applications in symbolic music processing.

Abstract

The recent explosion of generative AI-Music systems has raised numerous concerns over data copyright, licensing music from musicians, and the conflict between open-source AI and large prestige companies. Such issues highlight the need for publicly available, copyright-free musical data, in which there is a large shortage, particularly for symbolic music data. To alleviate this issue, we present PDMX: a large-scale open-source dataset of over 250K public domain MusicXML scores collected from the score-sharing forum MuseScore, making it the largest available copyright-free symbolic music dataset to our knowledge. PDMX additionally includes a wealth of both tag and user interaction metadata, allowing us to efficiently analyze the dataset and filter for high quality user-generated scores. Given the additional metadata afforded by our data collection process, we conduct multitrack music generation experiments evaluating how different representative subsets of PDMX lead to different behaviors in downstream models, and how user-rating statistics can be used as an effective measure of data quality. Examples can be found at https://pnlong.github.io/PDMX.demo/.
Paper Structure (15 sections, 3 figures, 3 tables)

This paper contains 15 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Dataset
  4. Parsing MuseScore Data
  5. Data Quality
  6. Deduplication
  7. Analysis
  8. Additional Features
  9. Experiments
  10. Experimental Setup and Metrics
  11. Results
  12. Objective Metrics
  13. Subjective Listening Test
  14. Conclusion
  15. Acknowledgements

Figures (3)

  • Figure 1: Pitch Class Entropy (PCE), Scale Consistency (SC), Groove Consistency (GC) vs. rating percentile in PDMX (specific ratings shown as vertical lines). Higher-rated songs seem to be more harmonically dynamic (higher PCE, lower SC), yet rating has little effect on rhythm.
  • Figure 2: Top-10 Genre distribution in PDMX. 67% of songs lack a genre tag. All genres besides the two most common, classical and folk music, display notably higher frequencies in the rated subsets.
  • Figure 3: Results from the subjective listening test. For each subset and rating axis, the mean opinion scores of the base and fine-tuned models are displayed on the left and right, respectively. The R$\cap$D (red) subset generally performs the best, and fine-tuning on $>$50% rated data improves richness and quality.