Temporal Grounding (Human Vocalization) on Audioset
6.6Acc (40ms)Gemini 2.5 Flash
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 2.5 FlashInference Mode=Zero-shot2026.02 | 6.6 | 12.7 | 2.3 | |
| Audio Flamingo 3Inference Mode=Zero-shot2026.02 | 5.8 | 10.2 | 1.69 | |
| Qwen2.5 7BInference Mode=Zero-shot2026.02 | 3.9 | 5.7 | 3.52 | |
| Qwen2.5 3BInference Mode=Zero-shot2026.02 | 3.6 | 4.5 | 4.46 | |
| Voxtral 24BInference Mode=Zero-shot2026.02 | 2.3 | 4.2 | 4.03 | |
| GPT-4o AudioInference Mode=Zero-shot2026.02 | 1.9 | 2.8 | 3.8 | |
| Voxtral 3BInference Mode=Zero-shot2026.02 | 1.2 | 2.2 | 3.74 |