Improving Source Separation by Explicitly Modeling Dependencies Between Sources

03/28/2022
by   Ethan Manilow, et al.
0

We propose a new method for training a supervised source separation system that aims to learn the interdependent relationships between all combinations of sources in a mixture. Rather than independently estimating each source from a mix, we reframe the source separation problem as an Orderless Neural Autoregressive Density Estimator (NADE), and estimate each source from both the mix and a random subset of the other sources. We adapt a standard source separation architecture, Demucs, with additional inputs for each individual source, in addition to the input mixture. We randomly mask these input sources during training so that the network learns the conditional dependencies between the sources. By pairing this training method with a block Gibbs sampling procedure at inference time, we demonstrate that the network can iteratively improve its separation performance by conditioning a source estimate on its earlier source estimates. Experiments on two source separation datasets show that training a Demucs model with an Orderless NADE approach and using Gibbs sampling (up to 512 steps) at inference time strongly outperforms a Demucs baseline that uses a standard regression loss and direct (one step) estimation of sources.

READ FULL TEXT

page 1

page 3

research
08/02/2018

Semi-blind source separation with multichannel variational autoencoder

This paper proposes a multichannel source separation method called the m...
research
08/14/2019

Interleaved Multitask Learning for Audio Source Separation with Independent Databases

Deep Neural Network-based source separation methods usually train indepe...
research
03/15/2012

Source Separation and Higher-Order Causal Analysis of MEG and EEG

Separation of the sources and analysis of their connectivity have been a...
research
06/21/2018

Towards Automated Single Channel Source Separation using Neural Networks

Many applications of single channel source separation (SCSS) including a...
research
04/07/2022

Heterogeneous Target Speech Separation

We introduce a new paradigm for single-channel target source separation ...
research
11/22/2022

Latent Iterative Refinement for Modular Source Separation

Traditional source separation approaches train deep neural network model...
research
03/27/2020

Separating Varying Numbers of Sources with Auxiliary Autoencoding Loss

Many recent source separation systems are designed to separate a fixed n...

Please sign up or login with your details

Forgot password? Click here to reset