# Few-Shot Musical Source Separation

arXiv: https://arxiv.org/abs/2205.01273 · HTML: https://mlsanity.com/paper/2205.01273/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 May 2022.

## Built on by (most cited first, 9 of 9)

- [Separate Anything You Describe](https://mlsanity.com/paper/2308.05037/) (2023)
- [A Stem-Agnostic Single-Decoder System for Music Source Separation Beyond Four Stems](https://mlsanity.com/paper/2406.18747/) (2024)
- [LC-Protonets: Multi-Label Few-Shot Learning for World Music Audio Tagging](https://mlsanity.com/paper/2409.11264/) (2024)
- [User-guided Generative Source Separation](https://mlsanity.com/paper/2507.01339/) (2025)
- [An Ensemble Approach to Music Source Separation: A Comparative Analysis of Conventional and Hierarchical Stem Separation](https://mlsanity.com/paper/2410.20773/) (2024)
- [FlowSep 2: Self-Supervised Flow Matching for Language-Queried Audio Source Separation](https://mlsanity.com/paper/2608.22111/) (2026)
- [Learning Music Audio Representations With Limited Data](https://mlsanity.com/paper/2505.06042/) (2025)
- [Episode-specific Fine-tuning for Metric-based Few-shot Learners with Optimization-based Training](https://mlsanity.com/paper/2506.17499/) (2025)
- [A Conditioned UNet for Music Source Separation](https://mlsanity.com/paper/2512.15532/) (2025)

## Abstract (arXiv)

Deep learning-based approaches to musical source separation are often limited to the instrument classes that the models are trained on and do not generalize to separate unseen instruments. To address this, we propose a few-shot musical source separation paradigm. We condition a generic U-Net source separation model using few audio examples of the target instrument. We train a few-shot conditioning encoder jointly with the U-Net to encode the audio examples into a conditioning vector to configure the U-Net via feature-wise linear modulation (FiLM). We evaluate the trained models on real musical recordings in the MUSDB18 and MedleyDB datasets. We show that our proposed few-shot conditioning paradigm outperforms the baseline one-hot instrument-class conditioned model for both seen and unseen instruments. To extend the scope of our approach to a wider variety of real-world scenarios, we also experiment with different conditioning example characteristics, including examples from different recordings, with multiple sources, or negative conditioning examples.