MuSViT: A Foundation Vision Model for Sheet Music Representation
Authors: Carlos Penarrubia, Antonio Rios-Vila, Eliseo Fuentes-Martinez, Juan C. Martinez-Sevilla, Francisco J. Castellanos, María Alfaro-Contreras, Jorge Calvo-Zaragoza
Organizations: Pattern Recognition and Artificial Intelligence Group, University of Alicante, Spain
Abstract
Foundation models have transformed vision and language processing by providing rich, reusable representations that transfer across diverse tasks. Sheet music, as a visual encoding of musical language, lacks such a strong domain-specific backbone. We introduce MuSViT (Music Score Vision Transformer): the first foundation vision model for sheet music representation -- a ViT encoder pre-trained via Masked Autoencoders on 9.7 million pages from the IMSLP. To handle the complexity of real-world scores, we adopt a two-stage curriculum: a synthetic warm-up on typeset scores followed by large-scale training on the full IMSLP corpus. We evaluate MuSViT on four downstream tasks -- full-page and staff-level music score recognition, music symbol detection, and score difficulty classification -- under two scenarios: linear probing (frozen encoder) and fine-tuning. Under linear probing, MuSViT consistently outperforms modern vision encoders, revealing that general-purpose representations, regardless of scale, fall systematically short on the structured symbolic properties of musical notation. Under fine-tuning, MuSViT generally improves upon task-specific state-of-the-art methods. An additional embedding-transcription consistency analysis reveals that MuSViT encodes symbolic musical structure directly in its representation space -- unlike other encoders, whose embeddings do not correlate with music notation content. These results establish MuSViT as a foundation backbone for sheet music understanding.
We propose a novel pipeline, Legato 2, for extracting symbolic notation and semantic knowledge from images of sheet music. Legato 2 features the first large-scale neural model for optical music recognition (OMR) to operate sequentially on a system-by-system basis, following the horizontal lines of notation as they are read on the page, rather than treating the page as an undifferentiated image, enabling better scaling to arbitrarily long inputs. It is also the first OMR model capable of generating symbolic transcriptions that include embedded textual content, such as titles and annotations. The pipeline combines system-level segmentation with an autoregressive vision-LM to capture both local notation details and score structure. Across multiple datasets, Legato 2 consistently outperforms prior state of the art. We also show that symbolic transcriptions complement visual inputs for frontier language models, improving their interpretation of dense musical documents. Legato 2 establishes new state-of-the-art performance in both OMR and downstream sheet music understanding.
Humans recognize a musical passage even when it is shifted in time or transposed in pitch, indicating a notion of equivariance in the representation space. Our analysis, however, shows that standard music transformers map such time-shifted or pitch-transposed inputs onto uncorrelated representations: these models become progressively less equivariant as they scale in size or train longer. This suggests that in standard music transformers, additional model capacity is allocated to memorizing absolute patterns rather than capturing shared musical structures. In this paper, we propose the Equivariant Music Transformer (EMT), which enforces equivariance through self-distillation by jointly optimizing a next-token-prediction and an auxiliary equivariance regularization loss. We find that the additional equivariance loss acts as a beneficial regularizer, simultaneously improving next-token prediction and producing equivariant latent representations. Through both objective and subjective evaluations, EMT demonstrates superior equivariance and generative capability compared to data augmentation, feature engineering, and state-of-the-art (SOTA) baselines. More broadly, our findings reveal that standard language modeling methods alone do not capture music's translational symmetries, and dedicated inductive biases are required to produce better music representations. The code, weights and demos are available online.
Learning the violin is harder than it looks. Unlike piano keys or guitar frets, the violin neck has no markings at all, so a beginner cannot tell by looking where to place each finger. MusicSynth is an open-source web tool that tries to fix that: user uploads a photo of any violin sheet music (or a digital score file), and the system automatically produces a video showing a violin fingerboard with each note highlighted at the right moment -- no software to install, no manual note entry required. The system connects three existing open-source tools into one pipeline: an optical music recognition (OMR) library reads the notes from the uploaded image, a MusicXML parser extracts timing information from digital scores, and a video renderer draws the fingerboard frame by frame. The only part built from scratch is the lookup table that maps each musical note to a string and finger position on the violin. Tested across 110 public-domain violin scores, MusicSynth correctly identified 91.2,% of notes in clean printed music and assigned the right finger position 99.1,% of the time when given a digital score file. To the author's knowledge, no freely available tool currently turns a sheet music image into an animated violin fingerboard tutorial automatically and in a single browser-based step.