BMdataset: A Musicologically Curated LilyPond Dataset

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Spanio, Matteo, Guler, Ilay, Rodà, Antonio
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915933500473344
author Spanio, Matteo
Guler, Ilay
Rodà, Antonio
author_facet Spanio, Matteo
Guler, Ilay
Rodà, Antonio
contents Symbolic music research has relied almost exclusively on MIDI-based datasets; text-based engraving formats such as LilyPond remain unexplored for music understanding. We present BMdataset, a musicologically curated dataset of 393 LilyPond scores (2,646 movements) transcribed by experts directly from original Baroque manuscripts, with metadata covering composer, musical form, instrumentation, and sectional attributes. Building on this resource, we introduce LilyBERT (weights can be found at https://huggingface.co/csc-unipd/lilybert), a CodeBERT-based encoder adapted to symbolic music through vocabulary extension with 115 LilyPond-specific tokens and masked language model pre-training. Linear probing on the out-of-domain Mutopia corpus shows that, despite its modest size (~90M tokens), fine-tuning on BMdataset alone outperforms continuous pre-training on the full PDMX corpus (~15B tokens) for both composer and style classification, demonstrating that small, expertly curated datasets can be more effective than large, noisy corpora for music understanding. Combining broad pre-training with domain-specific fine-tuning yields the best results overall (84.3% composer accuracy), confirming that the two data regimes are complementary. We release the dataset, tokenizer, and model to establish a baseline for representation learning on LilyPond.
format Preprint
id arxiv_https___arxiv_org_abs_2604_10628
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BMdataset: A Musicologically Curated LilyPond Dataset
Spanio, Matteo
Guler, Ilay
Rodà, Antonio
Sound
Computation and Language
Information Retrieval
68T07
H.5.5; I.2.7; J.5
Symbolic music research has relied almost exclusively on MIDI-based datasets; text-based engraving formats such as LilyPond remain unexplored for music understanding. We present BMdataset, a musicologically curated dataset of 393 LilyPond scores (2,646 movements) transcribed by experts directly from original Baroque manuscripts, with metadata covering composer, musical form, instrumentation, and sectional attributes. Building on this resource, we introduce LilyBERT (weights can be found at https://huggingface.co/csc-unipd/lilybert), a CodeBERT-based encoder adapted to symbolic music through vocabulary extension with 115 LilyPond-specific tokens and masked language model pre-training. Linear probing on the out-of-domain Mutopia corpus shows that, despite its modest size (~90M tokens), fine-tuning on BMdataset alone outperforms continuous pre-training on the full PDMX corpus (~15B tokens) for both composer and style classification, demonstrating that small, expertly curated datasets can be more effective than large, noisy corpora for music understanding. Combining broad pre-training with domain-specific fine-tuning yields the best results overall (84.3% composer accuracy), confirming that the two data regimes are complementary. We release the dataset, tokenizer, and model to establish a baseline for representation learning on LilyPond.
title BMdataset: A Musicologically Curated LilyPond Dataset
topic Sound
Computation and Language
Information Retrieval
68T07
H.5.5; I.2.7; J.5
url https://arxiv.org/abs/2604.10628