YNote: A Novel Music Notation for Fine-Tuning LLMs in Music Generation

TL;DR

YNote introduces a simplified, fixed-format music notation system using only four characters, facilitating easier fine-tuning of large language models for music generation and improving coherence and style in generated music.

Contribution

The paper presents YNote, a novel minimalistic music notation system that simplifies existing formats, enabling more effective fine-tuning of LLMs for music generation tasks.

Findings

Fine-tuning GPT-2 on YNote data achieved high BLEU and ROUGE scores.

YNote allows coherent and stylistically relevant music generation from minimal prompts.

YNote offers a practical alternative to complex music notations for machine learning.

Abstract

The field of music generation using Large Language Models (LLMs) is evolving rapidly, yet existing music notation systems, such as MIDI, ABC Notation, and MusicXML, remain too complex for effective fine-tuning of LLMs. These formats are difficult for both machines and humans to interpret due to their variability and intricate structure. To address these challenges, we introduce YNote, a simplified music notation system that uses only four characters to represent a note and its pitch. YNote's fixed format ensures consistency, making it easy to read and more suitable for fine-tuning LLMs. In our experiments, we fine-tuned GPT-2 (124M) on a YNote-encoded dataset and achieved BLEU and ROUGE scores of 0.883 and 0.766, respectively. With just two notes as prompts, the model was able to generate coherent and stylistically relevant music. We believe YNote offers a practical alternative to existing music notations for machine learning applications and has the potential to significantly enhance the quality of music generation using LLMs.