Generating Piano Music with Transformers: A Comparative Study of Scale, Data, and Metrics

Although a variety of transformers have been proposed for symbolic music generation in recent years, there is still little comprehensive study on how specific design choices affect the quality of the generated music. In this work, we systematically compare different datasets, model architectures, model sizes, and training strategies for the task of symbolic piano music generation. To support model development and evaluation, we examine a range of quantitative metrics and analyze how well they correlate with human judgment collected through listening studies. Our best-performing model, a 950M-parameter transformer trained on 80K MIDI files from diverse genres, produces outputs that are often rated as human-composed in a Turing-style listening survey.

• Publication year

  2025

• Venue

  arXiv.org

• Publication date

  2025-11-10

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.48550/arXiv.2511.07268arXiv 2511.07268
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Aria-MIDI: A Dataset of Piano MIDI Files for Symbolic Music Modeling

  2025influential reference
• Moonbeam: A MIDI Foundation Model Using Both Absolute and Relative Music Attributes

  2025cited by this paper
• Frechet Music Distance: A Metric For Generative Symbolic Music Evaluation

  2024cited by this paper
• PiJAMA: Piano Jazz with Automatic MIDI Annotations

  2023cited by this paper
• FIGARO: Controllable Music Generation using Learned and Expert Features

  2023cited by this paper
• ATEPP: A Dataset of Automatically Transcribed Expressive Piano Performance

  2022cited by this paper
• Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation

  2022cited by this paper
• A Domain-Knowledge-Inspired Music Embedding Space and a Novel Attention Mechanism for Symbolic Music Modeling

  2022cited by this paper
• A transformer generative adversarial network for multi-track music generation

  2021cited by this paper
• GiantMIDI-Piano: A large-scale MIDI dataset for classical piano music

  2020cited by this paper
• Pop Music Transformer: Beat-based Modeling and Generation of Expressive Pop Piano Compositions

  2020cited by this paper
• Computer-Generated Music for Tabletop Role-Playing Games

  2020cited by this paper
• On the evaluation of generative models in music

  2018cited by this paper
• Music Transformer: Generating Music with Long-Term Structure

  2018cited by this paper
• Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset

  2018influential reference
• This time with feeling: learning expressive musical performance

  2018cited by this paper
• MidiNet: A Convolutional Generative Adversarial Network for Symbolic-Domain Music Generation

  2017cited by this paper
• Learning-Based Methods for Comparing Sequences, with Applications to Audio-to-MIDI Alignment and Matching

  2016cited by this paper
• Computing Machinery and Intelligence

  1950cited by this paper

• No citing papers are available for this paper.