Vocal Event Detection on WESR-Bench Macro avg. 1.0
68.6PrecisionWESR-Whisper
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| WESR-WhisperMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 68.6 | 64.1 | 65.7 | |
| WESR-KimiMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 67.7 | 62.8 | 64.3 | |
| WESR-QwenMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 67.5 | 65 | 65.4 | |
| Gemini-2.5-ProPrompting=Few-shot2026.01 | 65 | 52.5 | 53.7 | |
| Gemini-3-ProPrompting=Few-shot2026.01 | 63.4 | 57.5 | 58 | |
| Qwen3-OmniPrompting=Few-shot2026.01 | 55.2 | 35.7 | 41 | |
| Kimi-AudioPrompting=Few-shot2026.01 | 29.3 | 14.7 | 16.6 |