Baseline dcase20 task 4

Nicolas Turpault, Romain SerizelZenodo 2020 · DOI

Abstract

Weights of the baseline models of dcase 20 task 4. The baseline model without sound separation is trained using 1 channel. The baseline model with sound separation is trained using 5 channels (1 for mixture and 4 for separated sounds).
BibTeX
@inproceedings{turpault2020baseline,
  author = {Nicolas Turpault and Romain Serizel},
  title = {{Baseline dcase20 task 4}},
  booktitle = {Zenodo},
  year = {2020},
  doi = {10.5281/zenodo.3726376},
}

Loading related content...

Figure