Picture for Jonathan Le Roux

Jonathan Le Roux

End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction

Apr 26, 2018
Zhong-Qiu Wang, Jonathan Le Roux, DeLiang Wang, John R. Hershey

Figure 1 for End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction

Figure 2 for End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction

Figure 3 for End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction

* Submitted to Interspeech 2018

Access Paper or Ask Questions

Deep Clustering and Conventional Networks for Music Separation: Stronger Together

Jun 15, 2017
Yi Luo, Zhuo Chen, John R. Hershey, Jonathan Le Roux, Nima Mesgarani

Figure 1 for Deep Clustering and Conventional Networks for Music Separation: Stronger Together

Figure 2 for Deep Clustering and Conventional Networks for Music Separation: Stronger Together

Figure 3 for Deep Clustering and Conventional Networks for Music Separation: Stronger Together

* Published in ICASSP 2017

Access Paper or Ask Questions

Full-Capacity Unitary Recurrent Neural Networks

Oct 31, 2016
Scott Wisdom, Thomas Powers, John R. Hershey, Jonathan Le Roux, Les Atlas

* 9 pages, to appear in NIPS

Access Paper or Ask Questions

Deep clustering: Discriminative embeddings for segmentation and separation

Aug 18, 2015
John R. Hershey, Zhuo Chen, Jonathan Le Roux, Shinji Watanabe

Figure 1 for Deep clustering: Discriminative embeddings for segmentation and separation

Figure 2 for Deep clustering: Discriminative embeddings for segmentation and separation

Figure 3 for Deep clustering: Discriminative embeddings for segmentation and separation

* Originally submitted on June 5, 2015

Access Paper or Ask Questions

Deep Unfolding: Model-Based Inspiration of Novel Deep Architectures

Nov 20, 2014
John R. Hershey, Jonathan Le Roux, Felix Weninger

Figure 1 for Deep Unfolding: Model-Based Inspiration of Novel Deep Architectures

Figure 2 for Deep Unfolding: Model-Based Inspiration of Novel Deep Architectures

* Added sections on reducing belief propagation to network activation functions, and on conversion between conventional network parameters and BP potentials for binary MRFs. Some bugs and typos were also fixed, and notation made a bit clearer

Access Paper or Ask Questions