Speech Recognition on LongSpeech
11WERUpper-bound
Evaluation results
| Method | Links | |
|---|---|---|
| Upper-boundcompression=uncompressed2026.02 | 11 | |
| Speech-XLcompression=SST mechanism2026.02 | 11.4 | |
| Whispermode=zero-shot2026.02 | 17 | |
| Voxtralmode=zero-shot2026.02 | 27.5 | |
| Qwen2-Audiomode=zero-shot2026.02 | 29.9 | |
| AudioFlamingo3mode=zero-shot2026.02 | 34.5 | |
| DashengLMmode=zero-shot2026.02 | 35.5 | |
| Kimi-audiomode=zero-shot2026.02 | 51.2 |