End-to-end Non-Negative Autoencoders for Sound Source Separation

TL;DR

This paper introduces end-to-end non-negative autoencoders for sound source separation, combining the modularity of NMF with neural network flexibility, achieving competitive results with easier adaptability.

Contribution

It develops a novel end-to-end non-negative autoencoder framework that unifies NMF's modularity with neural network capabilities for source separation.

Findings

Achieves separation performance comparable to discriminative models.

Retains modularity and adaptability of NMF.

Demonstrates flexibility of neural network-based approaches.

Abstract

Discriminative models for source separation have recently been shown to produce impressive results. However, when operating on sources outside of the training set, these models can not perform as well and are cumbersome to update. Classical methods like Non-negative Matrix Factorization (NMF) provide modular approaches to source separation that can be easily updated to adapt to new mixture scenarios. In this paper, we generalize NMF to develop end-to-end non-negative auto-encoders and demonstrate how they can be used for source separation. Our experiments indicate that these models deliver comparable separation performance to discriminative approaches, while retaining the modularity of NMF and the modeling flexibility of neural networks.