Speech Emotion Recognition on MELD-ST JPN (test)
51.58SER Weighted F1 ScoreERM-MinMaxGAP
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ERM-MinMaxGAPModality=Multimodal (Speech + Transcription)2026.03 | 51.58 | 52.19 | 8.13 | 2.97 | 1.22 | 1.44 | 3.44 | |
| gpt4o-mini-audioModality=Multimodal (Speech + Transcription)2026.03 | 48.56 | 47.71 | 7.77 | 2.69 | 2.99 | 3.75 | 4.3 | |
| ERM-MinMaxGAPModality=Unimodal (Speech only)2026.03 | 48.19 | 51.04 | 7.23 | 3.21 | 2.68 | 3.22 | 4.09 | |
| Voxtral-Mini-3BModality=Multimodal (Speech + Transcription)2026.03 | 47.91 | 54.06 | 7.01 | 2.12 | 0.81 | 2.03 | 2.99 | |
| Voxtral-Mini-3BModality=Unimodal (Speech only)2026.03 | 46.63 | 49.17 | 8.27 | 2.96 | 0.2 | 1.39 | 3.21 | |
| gpt4o-mini-audioModality=Unimodal (Speech only)2026.03 | 41.45 | 40.1 | 6.89 | 1.24 | 0.11 | 2.12 | 2.59 | |
| Qwen2-AudioModality=Unimodal (Speech only)2026.03 | 27.64 | 27.5 | 9.07 | 2.88 | 3.56 | 7.19 | 5.68 | |
| kimi-audio-7bModality=Multimodal (Speech + Transcription)2026.03 | 26.76 | 31.15 | 4.95 | 3.32 | 2.52 | 0.82 | 2.9 | |
| Ultravox-0.4Modality=Multimodal (Speech + Transcription)2026.03 | 25.66 | 25 | 8.12 | 2.52 | 3.99 | 1.77 | 4.1 | |
| Qwen2-AudioModality=Multimodal (Speech + Transcription)2026.03 | 25.56 | 23.85 | 3.61 | 1.95 | 3.03 | 2.29 | 2.72 | |
| kimi-audio-7bModality=Unimodal (Speech only)2026.03 | 23.91 | 26.67 | 5.5 | 4.18 | 5.72 | 1.5 | 4.23 | |
| Ultravox-0.4Modality=Unimodal (Speech only)2026.03 | 18.52 | 17.81 | 2.01 | 0.74 | 3.25 | 1.77 | 1.94 |