Skip to content

Latest commit

 

History

History
38 lines (25 loc) · 2.36 KB

description.md

File metadata and controls

38 lines (25 loc) · 2.36 KB
Raw

HEL1

Manal Helal, Dominic Ward, Mark Plumply CVSSP, Surrey University, Guildford, UK mhelal@cse.unsw.edu.au dominic.ward@surrey.ac.uk m.plumbley@surrey.ac.uk

Additional Info

  • is_blind: no
  • additional_training_data: no

Supplementary Material

  • Code:
  • Demos:

Method

This is tensor flow 5 layer RNN trained neural networks using stft magnitudes and soft time masking.

I will upload the source code, once I clean it up, and will write a paper with more details soon. It is based on a previous tensor flow implementation for 2 sources from ikala dataset: https://github.com/andabi/music-source-separation

extended to 4 sources from the musDB 2018 dataset, and scipy for feature extraction, and museval for evaluation.

I am working on more methods to submit soon, will write a paper then will clean up the code and upload it, and upload the demo files. After easter break will upload the demo files and update this file.

References

  • Weninger, F., Hershey, J. R., Roux, J. L., and Schuller, B., “Discriminatively trained recurrent neural networks for single-channel speech separation,” in Proc. GlobalSIP, 2014.
  • P.-S. Huang, M. Kim, M. Hasegawa-Johnson, and P. Smaragdis, “Singing-Voice separation from monaural recordings using deep recurrent neural networks,” in Proc. ISMIR, 2014, pp. 477–482. section 3-2
  • Weninger, F., Hershey, J. R., Roux, J. L., and Schuller, B., “Discriminatively trained recurrent neural networks for single-channel speech separation,” in Proc. GlobalSIP, 2014."
  • H. P. S., M. Kim, M. Hasegawa-Johnson, and P. Smaragdis, “Joint optimization of masks and deep recurrent neural networks for monaural source separation,” IEEE/ACM Trans. Audio Speech and Language Processing, vol. 23, no. 12, pp. 2136 – 2147, 2015.
  • P.-S. Huang, M. Kim, M. Hasegawa-Johnson, and P. Smaragdis, “Singing-Voice separation from monaural recordings using deep recurrent neural networks,” in Proc. ISMIR, 2014, pp. 477–482.
  • Emad M. Grais, Gerard Roma, Andrew J. R. Simpson, and Mark D. Plumbley, "Two Stage Single Channel Audio Source Separation using Deep Neural Networks"