# Demucs: Deep Extractor for Music Sources with extra unlabeled data remixed

arXiv: https://arxiv.org/abs/1909.01174 · HTML: https://mlsanity.com/paper/1909.01174/

## Key facts

- Cited by 54 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Sep 2019.

## Built on by (most cited first, 30 of 54)

- [Music Source Separation with Band-split RNN](https://mlsanity.com/paper/2209.15174/) (2022)
- [Separate Anything You Describe](https://mlsanity.com/paper/2308.05037/) (2023)
- [IndexTTS: An Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System](https://mlsanity.com/paper/2502.05512/) (2025)
- [Zero-shot Audio Source Separation through Query-based Learning from Weakly-labeled Data](https://mlsanity.com/paper/2112.07891/) (2021)
- [MusicAgent: An AI Agent for Music Understanding and Generation with Large Language Models](https://mlsanity.com/paper/2310.11954/) (2023)
- [Content-based Controls For Music Large Language Modeling](https://mlsanity.com/paper/2310.17162/) (2023)
- [LLaSE-G1: Incentivizing Generalization Capability for LLaMA-based Speech Enhancement](https://mlsanity.com/paper/2503.00493/) (2025)
- [Analyzable Chain-of-Musical-Thought Prompting for High-Fidelity Music Generation](https://mlsanity.com/paper/2503.19611/) (2025)
- [ReZero: Region-customizable Sound Extraction](https://mlsanity.com/paper/2308.16892/) (2023)
- [MSAVBench: Towards Comprehensive and Reliable Evaluation of Multi-Shot Audio-Video Generation](https://mlsanity.com/paper/2605.20183/) (2026)
- [CWS-PResUNet: Music Source Separation with Channel-wise Subband Phase-aware ResUNet](https://mlsanity.com/paper/2112.04685/) (2021)
- [OmniSep: Unified Omni-Modality Sound Separation with Query-Mixup](https://mlsanity.com/paper/2410.21269/) (2024)
- [A Stem-Agnostic Single-Decoder System for Music Source Separation Beyond Four Stems](https://mlsanity.com/paper/2406.18747/) (2024)
- [FastSAG: Towards Fast Non-Autoregressive Singing Accompaniment Generation](https://mlsanity.com/paper/2405.07682/) (2024)
- [SaMoye: Zero-shot Singing Voice Conversion Model Based on Feature Disentanglement and Enhancement](https://mlsanity.com/paper/2407.07728/) (2024)
- [CycleDRUMS: Automatic Drum Arrangement For Bass Lines Using CycleGAN](https://mlsanity.com/paper/2104.00353/) (2021)
- [Transfer Learning with Jukebox for Music Source Separation](https://mlsanity.com/paper/2111.14200/) (2021)
- [Unsupervised Music Source Separation Using Differentiable Parametric Source Models](https://mlsanity.com/paper/2201.09592/) (2022)
- [MUSE: Flexible Voiceprint Receptive Fields and Multi-Path Fusion Enhanced Taylor Transformer for U-Net-based Speech Enhancement](https://mlsanity.com/paper/2406.04589/) (2024)
- [Adaptive Slimming for Scalable and Efficient Speech Enhancement](https://mlsanity.com/paper/2507.04879/) (2025)
- [UniFlow: Unifying Speech Front-End Tasks via Continuous Generative Modeling](https://mlsanity.com/paper/2508.07558/) (2025)
- [SAO-Instruct: Free-form Audio Editing using Natural Language Instructions](https://mlsanity.com/paper/2510.22795/) (2025)
- [Audiovisual Singing Voice Separation](https://mlsanity.com/paper/2107.00231/) (2021)
- [Looking Similar, Sounding Different: Leveraging Counterfactual Cross-Modal Pairs for Audiovisual Representation Learning](https://mlsanity.com/paper/2304.05600/) (2023)
- [The Radio-Frequency Transformer for Signal Separation](https://mlsanity.com/paper/2603.09201/) (2026)
- [OmniHuman: A Large-scale Dataset and Benchmark for Human-Centric Video Generation](https://mlsanity.com/paper/2604.18326/) (2026)
- [On the Design of Deep Priors for Unsupervised Audio Restoration](https://mlsanity.com/paper/2104.07161/) (2021)
- [Improved singing voice separation with chromagram-based pitch-aware remixing](https://mlsanity.com/paper/2203.15092/) (2022)
- [Subtractive Training for Music Stem Insertion using Latent Diffusion Models](https://mlsanity.com/paper/2406.19328/) (2024)
- [U-Mamba-Net: A highly efficient Mamba-based U-net style network for noisy and reverberant speech separation](https://mlsanity.com/paper/2412.18217/) (2024)

## Abstract (arXiv)

We study the problem of source separation for music using deep learning with four known sources: drums, bass, vocals and other accompaniments. State-of-the-art approaches predict soft masks over mixture spectrograms while methods working on the waveform are lagging behind as measured on the standard MusDB benchmark. Our contribution is two fold. (i) We introduce a simple convolutional and recurrent model that outperforms the state-of-the-art model on waveforms, that is, Wave-U-Net, by 1.6 points of SDR (signal to distortion ratio). (ii) We propose a new scheme to leverage unlabeled music. We train a first model to extract parts with at least one source silent in unlabeled tracks, for instance without bass. We remix this extract with a bass line taken from the supervised dataset to form a new weakly supervised training example. Combining our architecture and scheme, we show that waveform methods can play in the same ballpark as spectrogram ones.