Codecformer-EL replaces the waveform loss in codec-based speech separation with an MSE loss on frozen codec embeddings, cutting training cost by about half and improving perceptual metrics on WSJ0-2mix.
TF- GridNet: Integrating full- and sub-band modeling for speech separation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Speech Separation using Neural Audio Codecs with Embedding Loss
Codecformer-EL replaces the waveform loss in codec-based speech separation with an MSE loss on frozen codec embeddings, cutting training cost by about half and improving perceptual metrics on WSJ0-2mix.