Program
Monaural Source Separation
Wed-2-4-4 Asteroid : the PyTorch-based audio source separation toolkit for researchers
Manuel Pariente(Université de Lorraine, CNRS, Inria, LORIA, F-54000 Nancy, France), Samuele Cornell(Università Politecnica delle Marche), Joris Cosentino(Inria), Sunit Sivasankaran(INRIA), Efthymios Tzinis(University of Illinois at Urbana-Champaign), Jens Heitkaemper(Paderborn University), Michel Olvera(Université de Lorraine), Fabian-Robert Stöter(Inria and LIRMM, University of Montpellier), Mathieu Hu(Inria), Juan M. Martín-Doñas(University of Granada), David Ditter(University of Hamburg), Ariel Frank(Technion - Israel Institute of Technology), Antoine Deleforge(INRIA) and Emmanuel Vincent(INRIA)
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 https://github.com/mpariente/asteroid.
Calls
Program