Proposes multichannel Itakura-Saito divergence losses for training DNN mask estimators used in mask-based beamforming for supervised speech source separation.
(3) [10] and by multi- channel loss functions were compared in speaker-independent multi-talker separation by the mask-based beamforming
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Multichannel Loss Function for Supervised Speech Source Separation by Mask-based Beamforming
Proposes multichannel Itakura-Saito divergence losses for training DNN mask estimators used in mask-based beamforming for supervised speech source separation.