Speaker Description on LibriTTS + DEMAND mixtures Foreground
99.7Gender AccQwen2-Audio
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen2-AudioSystem Setup=Upper Bound: Oracle Speaker + Auditory LLM2025.02 | 99.7 | 94.6 | 91.6 | |
| AAD-LLMSystem Setup=Intention-Informed AAD-LLM2025.02 | 99.4 | 90.9 | 88 | |
| Qwen2-AudioSystem Setup=Proposed Baseline: Extracted Speaker + Auditory LLM2025.02 | 99.1 | 89.5 | 87.4 | |
| Qwen2-AudioSystem Setup=Lower Bound: Random Speaker + Auditory LLM2025.02 | 75.9 | 63.4 | 73.1 | |
| LTU-ASSystem Setup=Auditory LLM without Intention2025.02 | 73.7 | 47.1 | 26.7 | |
| SALMONNSystem Setup=Auditory LLM without Intention2025.02 | 73.2 | 33.5 | 71.4 | |
| Qwen2-AudioSystem Setup=Auditory LLM without Intention2025.02 | 71.9 | 34.4 | 59.7 | |
| GAMASystem Setup=Auditory LLM without Intention2025.02 | 60.2 | 32.6 | 52.9 | |
| WavLLMSystem Setup=Auditory LLM without Intention2025.02 | 53.6 | 31.9 | 53.5 | |
| Qwen-AudioSystem Setup=Auditory LLM without Intention2025.02 | 48.2 | 37.5 | 8.5 |