End-to-end Non-Negative Autoencoders for Sound Source Separation
This work addresses the need for adaptable source separation methods that can handle new sources without extensive retraining, offering a modular alternative to discriminative approaches.
The paper tackled the problem of sound source separation by generalizing Non-negative Matrix Factorization (NMF) into end-to-end non-negative autoencoders, achieving comparable performance to discriminative models while maintaining modularity and flexibility.
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.