Technical Program

Click on the icon to view the manuscript on IEEE XPlore in the IEEE ICASSP 2020 Open Preview.

AUD-L2: Deep Learning for Source Separation

Session Type: Lecture
Time: Tuesday, 5 May, 16:30 - 18:30
Location: On-Demand
Virtual Session: View on Virtual Platform
Session Chair: Minje Kim, Indiana University Bloomington
 
 AUD-L2.1: TWO-STEP SOUND SOURCE SEPARATION: TRAINING ON LEARNED LATENT TARGETS
         Efthymios Tzinis; University of Illinois at Urbana-Champaign
         Shrikant Venkataramani; University of Illinois at Urbana-Champaign
         Zhepei Wang; University of Illinois at Urbana-Champaign
         Cem Subakan; Mila-Quebec Artificial Intelligence Institute
         Paris Smaragdis; University of Illinois at Urbana-Champaign, Adobe Research
 
 AUD-L2.2: A MULTI-PHASE GAMMATONE FILTERBANK FOR SPEECH SEPARATION VIA TASNET
         David Ditter; Universit├Ąt Hamburg
         Timo Gerkmann; Universit├Ąt Hamburg
 
 AUD-L2.3: IMPROVING VOICE SEPARATION BY INCORPORATING END-TO-END SPEECH RECOGNITION
         Naoya Takahashi; Sony Corporation
         Mayank Singh; Indian Institute of Technology Bombay
         Sakya Basak; Indian Institute of Science
         Parthasaarathy Sudarsanam; Sony India Software Centre
         Sriram Ganapathy; Indian Institute of Science
         Yuki Mitsufuji; Sony Corporation
 
 AUD-L2.4: DUAL-PATH RNN: EFFICIENT LONG SEQUENCE MODELING FOR TIME-DOMAIN SINGLE-CHANNEL SPEECH SEPARATION
         Yi Luo; Columbia University
         Zhuo Chen; Microsoft
         Takuya Yoshioka; Microsoft
 
 AUD-L2.5: CONTROLLING THE PERCEIVED SOUND QUALITY FOR DIALOGUE ENHANCEMENT WITH DEEP LEARNING
         Christian Uhle; Fraunhofer Institute for Integrated Circuits IIS
         Matteo Torcoli; Fraunhofer Institute for Integrated Circuits IIS
         Jouni Paulus; Fraunhofer Institute for Integrated Circuits IIS
 
 AUD-L2.6: UNSUPERVISED TRAINING FOR DEEP SPEECH SOURCE SEPARATION WITH KULLBACK-LEIBLER DIVERGENCE BASED PROBABILISTIC LOSS FUNCTION
         Masahito Togami; LINE Corporation
         Yoshiki Masuyama; Waseda University
         Tatsuya Komatsu; LINE Corporation
         Yu Nakagome; Waseda University