AudioBench - CoVoST2 TA-EN BLEU: leaderboard
AudioBench CoVoST2 TA-EN BLEU speech-translation BLEU slice measuring translation quality between the named source and target languages.
Metric: Mean Score (%). Source: github.com. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | SALMONN-7B | 36.49 | #1293 |
| 2 | SeaLLMs-Audio-7B | 6.48 | #1018 |
| 3 | Phi-4 Multimodal Instruct | 5.31 | #906 |
| 4 | Qwen2-Audio-7B-Instruct | 4.43 | #880 |
| 5 | Whisper Large-v3 | 2.45 | #1271 |
| 6 | Qwen-Audio-Chat | 1.7 | #1244 |
Interactive version: theaggregate.ai/benchmark?slug=audiobench-covost2-ta-en-bleu · How It Works · Data refreshed daily, snapshot 2026-10-20.