Asteroid: the PyTorch-based audio source separation toolkit for researchers
This toolkit addresses the need for accessible and reproducible tools for researchers in audio source separation, though it is incremental as it builds on existing methods.
The paper introduces Asteroid, a PyTorch-based toolkit for audio source separation, providing neural building blocks and Kaldi-style recipes to improve reproducibility, and shows that its implementations achieve results at least on par with reference papers.
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 https://github.com/mpariente/asteroid .