Vocal Event Detection on WESR-Bench 1.0
0.75PrecisionWESR-Whisper
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| WESR-WhisperMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 0.75 | 0.71 | 0.73 | |
| WESR-KimiMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 0.736 | 0.719 | 0.728 | |
| WESR-QwenMode=Fine-tuned, Training Data=WESR-Train corpus2026.01 | 0.735 | 0.737 | 0.736 | |
| Gemini-2.5-ProPrompting=Few-shot2026.01 | 0.682 | 0.483 | 0.566 | |
| Gemini-3-ProPrompting=Few-shot2026.01 | 0.665 | 0.57 | 0.614 | |
| Qwen3-OmniPrompting=Few-shot2026.01 | 0.58 | 0.327 | 0.418 | |
| Kimi-AudioPrompting=Few-shot2026.01 | 0.262 | 0.058 | 0.094 |