MGE-LDM: Joint Latent Diffusion for Simultaneous Music Generation and Source Extraction

TL;DR

MGE-LDM introduces a unified latent diffusion model capable of simultaneous music generation, source imputation, and flexible source separation, trained across diverse datasets without fixed instrument categories.

Contribution

The paper presents a novel joint latent diffusion framework that handles multiple music tasks and sources in a class-agnostic manner, unlike prior fixed-category models.

Findings

Supports complete and partial music generation.

Enables text-conditioned source extraction.

Trained on heterogeneous multi-track datasets.

Abstract

We present MGE-LDM, a unified latent diffusion framework for simultaneous music generation, source imputation, and query-driven source separation. Unlike prior approaches constrained to fixed instrument classes, MGE-LDM learns a joint distribution over full mixtures, submixtures, and individual stems within a single compact latent diffusion model. At inference, MGE-LDM enables (1) complete mixture generation, (2) partial generation (i.e., source imputation), and (3) text-conditioned extraction of arbitrary sources. By formulating both separation and imputation as conditional inpainting tasks in the latent space, our approach supports flexible, class-agnostic manipulation of arbitrary instrument sources. Notably, MGE-LDM can be trained jointly across heterogeneous multi-track datasets (e.g., Slakh2100, MUSDB18, MoisesDB) without relying on predefined instrument categories. Audio samples are available at our project page: https://yoongi43.github.io/MGELDM_Samples/.