Benchmark explorer.
Open ASR · URGENT2024 WEREnglish short-form; normalization unversioned · Lower is better
Source ↗Method
English short-form word error rate after source text normalization. Lower is better; insertions can raise WER above 100%. Leaderboard publication; individual runner and verification status are not established unless separately attributed.
| # | Model | URGENT2024 WER | Configuration | API rate / unit | Source |
|---|---|---|---|---|---|
| 51 | microsoft/VibeVoice-ASR-HFmicrosoft | 19.5% WER | microsoft/VibeVoice-ASR-HFLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 52 | nvidia/nemotron-3.5-asr-streaming-0.6bnvidia | 20.14% WER | nvidia/nemotron-3.5-asr-streaming-0.6bLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 53 | abr-ai/niagara-84m-batch.enabr-ai | 20.16% WER | abr-ai/niagara-84m-batch.enLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 54 | abr-ai/niagara-38m-batch.enabr-ai | 22.88% WER | abr-ai/niagara-38m-batch.enLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 55 | abr-ai/niagara-19m-batch.enabr-ai | 25.4% WER | abr-ai/niagara-19m-batch.enLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 56 | google/gemma-4-E4B-itgoogle | 26.38% WER | google/gemma-4-E4B-itLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |
| 57 | abr-ai/niagara-9m-batch.enabr-ai | 29.47% WER | abr-ai/niagara-9m-batch.enLeaderboard entry | See provider quotes | Result ↗Oct 22, 2026 |