Showing eess.ASShow all
2 papers · 1 filter
eess.AS2026
VeRe-Flow: Guiding Flow Matching toward Clean Speech via Velocity Contrastive Regularization and Representation Alignment for Noise-Robust Bandwidth Expansion
Sujin Koo, Sangyoon Kim, Ji Sub Um +1
Noise-robust bandwidth expansion aims to reconstruct high-fidelity wideband speech from noisy low-resolution inputs. While flow matching has shown strong performance in speech gene…
eess.AS2026
How Far Do SSL Speech Models Listen for Tone? Temporal Focus of Tone Representation under Low-resource Transfer
Minu Kim, Ji Sub Um, Hoirin Kim
Lexical tone is central to many languages but remains underexplored in self-supervised learning (SSL) speech models, especially beyond Mandarin. We study four languages with comple…