Audio Understanding on MMAR (Component Scores)
45.24Speech ScoreUniAudio-Token
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| UniAudio-TokenBackbone=Qwen2.5-3B, Tuning strategy=tuned identically2026.05 | 45.24 | 43.64 | 40.29 | 45.8 | |
| StableTokenBackbone=Qwen2.5-3B, Tuning strategy=tuned identically2026.05 | 42.18 | 39.39 | 31.07 | 39.1 | |
| CosyVoice2Backbone=Qwen2.5-3B, Tuning strategy=tuned identically2026.05 | 41.5 | 35.76 | 30.58 | 38.1 | |
| WavTokenizerBackbone=Qwen2.5-3B, Tuning strategy=tuned identically2026.05 | 39.8 | 31.52 | 29.61 | 36.3 | |
| GLM-4-Voice-TokenizerBackbone=Qwen2.5-3B, Tuning strategy=tuned identically2026.05 | 39.46 | 40 | 36.89 | 40.1 |