Unifying Symbolic Music Arrangement: Track-Aware Reconstruction and Structured Tokenization

Ye Wang (National University of Singapore) · Longshen Ou (National University of Singapore) · Jingwei Zhao (NUS) · Ziyu Wang (New York University) · Gus Xia (New York University) · Qihao Liang (National University of Singapore) · Torin Hopkins (National University of Singapore)

any-to-any instrumentation transformationsautomatic multitrack music arrangementmodeling efficiencyperceptual evaluationsreinterpretationremi-zsegment-level reconstructionsimplificationstructured tokenizationstyle transfersymbolic music modelsymbolic music-to-music transformationtask-specific modelstoken-level disentangled contenttrack-wise modelingunconditional generation

We present a unified framework for automatic multitrack music arrangement that enables a single pre-trained symbolic music model to handle diverse arrangement scenarios, including reinterpretation, simplification, and additive generation. At its core is a segment-level reconstruction objective operating on token-level disentangled content and style, allowing for flexible any-to-any instrumentation transformations at inference time. To support track-wise modeling, we introduce REMI-z, a structured tokenization scheme for multitrack symbolic music that enhances modeling efficiency and effectiveness for both arrangement tasks and unconditional generation. Our method outperforms task-specific state-of-the-art models on representative tasks in different arrangement scenarios