# Evaluating Disentangled Representations for Controllable Music Generation

arXiv: https://arxiv.org/abs/2602.10058 · HTML: https://mlsanity.com/paper/2602.10058/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 25 references that are in the map.
- First posted on arXiv 10 Feb 2026.

## Builds on (most cited first, 21 of 21)

- [Simple and Controllable Music Generation](https://mlsanity.com/paper/2306.05284/) (2023)
- [MusicLM: Generating Music From Text](https://mlsanity.com/paper/2301.11325/) (2023)
- [Enabling Factorized Piano Music Modeling and Generation with the MAESTRO Dataset](https://mlsanity.com/paper/1810.12247/) (2018)
- [Stable Audio Open](https://mlsanity.com/paper/2407.14358/) (2024)
- [Isolating Sources of Disentanglement in Variational Autoencoders](https://mlsanity.com/paper/1802.04942/) (2018)
- [Disentangled Sequential Autoencoder](https://mlsanity.com/paper/1803.02991/) (2018)
- [Cutting Music Source Separation Some Slakh: A Dataset to Study the Impact of Training Data Quality and Quantity](https://mlsanity.com/paper/1909.08494/) (2019)
- [Music ControlNet: Multiple Time-varying Controls for Music Generation](https://mlsanity.com/paper/2311.07069/) (2023)
- [Codified audio language modeling learns useful representations for music information retrieval](https://mlsanity.com/paper/2107.05677/) (2021)
- [A Foundation Model for Music Informatics](https://mlsanity.com/paper/2311.03318/) (2023)
- [Joint Audio and Symbolic Conditioning for Temporally Controlled Text-to-Music Generation](https://mlsanity.com/paper/2406.10970/) (2024)
- [Unaligned Supervision For Automatic Music Transcription in The Wild](https://mlsanity.com/paper/2204.13668/) (2022)
- [Seed-Music: A Unified Framework for High Quality and Controlled Music Generation](https://mlsanity.com/paper/2409.09214/) (2024)
- [Learning Disentangled Representations of Timbre and Pitch for Musical Instrument Sounds Using Gaussian Mixture Variational Autoencoders](https://mlsanity.com/paper/1906.08152/) (2019)
- [Self-Supervised VQ-VAE for One-Shot Music Style Transfer](https://mlsanity.com/paper/2102.05749/) (2021)
- [Do Music Generation Models Encode Music Theory?](https://mlsanity.com/paper/2410.00872/) (2024)
- [Combining audio control and style transfer using latent diffusion](https://mlsanity.com/paper/2408.00196/) (2024)
- [Is Disentanglement enough? On Latent Representations for Controllable Music Generation](https://mlsanity.com/paper/2108.01450/) (2021)
- [Towards a Unified Representation Evaluation Framework Beyond Downstream Tasks](https://mlsanity.com/paper/2505.06224/) (2025)
- [Towards Robust Unsupervised Disentanglement of Sequential Data -- A Case Study Using Music Audio](https://mlsanity.com/paper/2205.05871/) (2022)
- [OMAR-RQ: Open Music Audio Representation Model Trained with Multi-Feature Masked Token Prediction](https://mlsanity.com/paper/2507.03482/) (2025)

## Built on by (most cited first, 1 of 1)

- [MusGU+: Toward a Musician-Centered Evaluation Framework and Discovery Tool for Generative Music AI](https://mlsanity.com/paper/2608.30940/) (2026)

## Abstract (arXiv)

Recent approaches in music generation rely on disentangled representations, often labeled as structure and timbre or local and global, to enable controllable synthesis. Yet the underlying properties of these embeddings remain underexplored. In this work, we evaluate such disentangled representations in a set of music audio models for controllable generation using a probing-based framework that goes beyond standard downstream tasks. The selected models reflect diverse unsupervised disentanglement strategies, including inductive biases, data augmentations, adversarial objectives, and staged training procedures. We further isolate specific strategies to analyze their effect. Our analysis spans four key axes: informativeness, equivariance, invariance, and disentanglement, which are assessed across datasets, tasks, and controlled transformations. Our findings reveal inconsistencies between intended and actual semantics of the embeddings, suggesting that current strategies fall short of producing truly disentangled representations, and prompting a re-examination of how controllability is approached in music generation.