Speech Emotion Recognition on MELD → IEMOCAP Cross-Domain
51.75Weighted F1VowelPrompt
Evaluation Results
| Method | Links | |
|---|---|---|
| VowelPromptEvaluation Regime=SFT & GRPO, Backbone=LLaMA-3-8B-Instruct2026.02 | 51.75 | |
| VowelPrompt2026.02 | 51.7 | |
| VowelPromptEvaluation Regime=Zero-Shot, Backbone=LLaMA-3-8B-Instruct2026.02 | 46.26 | |
| SpeechCueLLMEvaluation Regime=SFT & GRPO, Backbone=LLaMA-3-8B-Instruct2026.02 | 44.79 | |
| HuBERT-largeClassifier=MLP, Feature=Self-supervised embeddings2026.02 | 44.2 | |
| InstructERCEvaluation Regime=SFT & GRPO, Backbone=LLaMA-3-8B-Instruct2026.02 | 43.36 | |
| InstructERCEvaluation Regime=Zero-Shot, Backbone=LLaMA-3-8B-Instruct2026.02 | 42.68 | |
| SpeechCueLLMEvaluation Regime=Zero-Shot, Backbone=LLaMA-3-8B-Instruct2026.02 | 42.59 | |
| wav2vec-largeClassifier=MLP, Feature=Self-supervised embeddings2026.02 | 41.7 | |
| SALMONNEvaluation Regime=SFT & GRPO, Backbone=LLaMA-3-8B-Instruct2026.02 | 40.85 | |
| VowelPromptEvaluation Regime=SFT, Backbone=LLaMA-3-8B-Instruct2026.02 | 28.71 | |
| InstructERCEvaluation Regime=SFT, Backbone=LLaMA-3-8B-Instruct2026.02 | 25.49 | |
| SpeechCueLLMEvaluation Regime=SFT, Backbone=LLaMA-3-8B-Instruct2026.02 | 25.1 | |
| SALMONNEvaluation Regime=SFT, Backbone=LLaMA-3-8B-Instruct2026.02 | 23.65 |