Spatial-Magnifier neural network creates virtual microphone signals from real ones and uses the SARL framework to condition speech enhancement systems, nearly recovering full-array oracle performance.
Efficient audiovisual speech processing via mutud: Multimodal training and unimodal deployment
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Spatial-Magnifier: Spatial upsampling for multichannel speech enhancement
Spatial-Magnifier neural network creates virtual microphone signals from real ones and uses the SARL framework to condition speech enhancement systems, nearly recovering full-array oracle performance.