Decoupled Feature and Self-Knowledge Distillation for Speech Emotion Recognition
Speech Emotion Recognition seeks to capture emotional information from speech signals. While prior research has predominantly focused on the extraction and fusion of acoustic and end-to-end features, redundancy within these features often hinders model performance. In this work, we introduce DFSD-EM...
| Hauptverfasser: | , |
|---|---|
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
IEEE
2025-01-01
|
| Schriftenreihe: | IEEE Access |
| Schlagworte: | |
| Online Zugang: | https://ieeexplore.ieee.org/document/10891411/ |