Noise augmentation yields the largest gains for pathological speech enhancement while generative methods can degrade performance, though a gap to neurotypical speech remains.
The third ‘CHiME’ speech separation and recognition challenge: Dataset, task and base- lines
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
-
Data Augmentation for Pathological Speech Enhancement
Noise augmentation yields the largest gains for pathological speech enhancement while generative methods can degrade performance, though a gap to neurotypical speech remains.