Published September 4, 2017 | Version v1

Trained model for DSD100 stereo source separation using DeepConvSep

Authors/Creators

  • 1. Universitat Pompeu Fabra, Barcelona

Description

Trained model for DSD100 stereo source separation using the two channel DeepConvSep architecture (without ILD enhancing). It can be used for separation purposes following the documentation in the repository:

https://github.com/MTG/DeepConvSep

The specific two channel architecture is described in:

Erruz, Gerard (2017). "Binaural Source Separation with Convolutional Neural Networks". Master Thesis. Music Technology Group (Universitat Pompeu Fabra, Barcelona)

Files

Files (8.9 MB)

Name Size Download all
md5:40b0bbf354a002637aeecf9d5b080ebe
8.9 MB Download

Additional details

Related works

Is documented by
10.5281/zenodo.1095834 (DOI)

References

  • G. Erruz (2017). "Binaural Source Separation with Convolutional Neural Networks". Master Thesis. Music Technology Group (Universitat Pompeu Fabra, Barcelona) (to be published)
  • P. Chandna, M. Miron, J. Janer, and E. Gomez, "Monoaural audio source separation using deep convolutional neural networks" International Conference on Latent Variable Analysis and Signal Separation, 2017.