Audio Captioning on Song Describer (SD)
0.469SBERT SimilarityCaptioner
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CaptionerStyle=Cross2026.02 | 0.469 | — | |
| CaptionerStyle=Matched2026.02 | 0.468 | — | |
| MetadataStyle=Cross2026.02 | 0.462 | — | |
| MetadataStyle=Matched2026.02 | 0.454 | — | |
| Kimi-Audio-InstructDecoder Size=7B2025.08 | — | 44.63 | |
| MiDashengLMDecoder Size=7B2025.08 | — | 45.39 | |
| Qwen2.5-OmniDecoder Size=7B2025.08 | — | 45.31 |