Abstract
This paper describes Asteroid, the PyTorch-based audio source separation toolkit for researchers. Inspired by the most successful neural source separation systems, it provides all neural building blocks required to build such a system. To improve reproducibility, Kaldi-style recipes on common audio source separation datasets are also provided. This paper describes the software architecture of Asteroid and its most important features. By showing experimental results obtained with Asteroid's recipes, we show that our implementations are at least on par with most results reported in reference papers. The toolkit is publicly available at github.com/mpariente/asteroid.
Original language | English |
---|---|
Pages (from-to) | 2637-2641 |
Number of pages | 5 |
Journal | Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH |
Volume | 2020-October |
DOIs | |
State | Published - 2020 |
Externally published | Yes |
Event | 21st Annual Conference of the International Speech Communication Association, INTERSPEECH 2020 - Shanghai, China Duration: 25 Oct 2020 → 29 Oct 2020 |
Bibliographical note
Publisher Copyright:© 2020 International Speech Communication Association. All rights reserved.
Funding
Experiments presented in this paper were partially carried out using the Grid'5000 testbed, supported by a scientific interest group hosted by Inria and including CNRS, RENATER and several Universities as well as other organizations (see https://www.grid5000.fr). High Performance Computing resources were partially provided by the EX-PLOR centre hosted by the University de Lorraine.
Funders | Funder number |
---|---|
RENATER | |
University de Lorraine | |
Centre National de la Recherche Scientifique | |
several Universities |
Keywords
- End-to-end
- Open-source software
- Source separation
- Speech enhancement