Abstract
The sources separated by most single channel audio source separation techniques are usually distorted and each separated source contains residual signals from the other sources. To tackle this problem, we propose to enhance the separated sources to decrease the distortion and interference between the separated sources using deep neural networks (DNNs). Two different DNNs are used in this work. The first DNN is used to separate the sources from the mixed signal. The second DNN is used to enhance the separated signals. To consider the interactions between the separated sources, we propose to use a single DNN to enhance all the separated sources together. To reduce the residual signals of one source from the other separated sources (interference), we train the DNN for enhancement discriminatively to maximize the dissimilarity between the predicted sources. The experimental results show that using discriminative enhancement decreases the distortion and interference between the separated sources.
Original language | English |
---|---|
Title of host publication | Latent Variable Analysis and Signal Separation |
Editors | Petr Tichavský, Massoud Babaie-Zadeh, Olivier J.J. Michel, Nadège Thirion-Moreau |
Publisher | Springer, Cham |
Pages | 236-246 |
Number of pages | 11 |
ISBN (Electronic) | 9783319535470 |
ISBN (Print) | 9783319535463 |
DOIs | |
Publication status | Published - 15 Feb 2017 |
Externally published | Yes |
Event | 13th International Conference on Latent Variable Analysis and Signal Separation - Grenoble-Alpes University, Grenoble, France Duration: 21 Feb 2017 → 23 Feb 2017 Conference number: 13 http://www.lva-ica-2017.com/ (Link to Conference Website) |
Publication series
Name | Lecture Notes in Computer Science |
---|---|
Publisher | Springer |
Volume | 10169 |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | 13th International Conference on Latent Variable Analysis and Signal Separation |
---|---|
Abbreviated title | LVA/ICA 2017 |
Country | France |
City | Grenoble |
Period | 21/02/17 → 23/02/17 |
Internet address |
|