canary-v2
NVIDIA🇺🇸 USMajor lab·
Audioopen weights
No description yet — the model card or announcement hasn't been summarised.
- Modalities
- text → text
1 variant
| 1B | 1B | HF |
Sources
HuggingFaceBenchmarks
| Benchmark | Domain | Value | Evidence |
|---|---|---|---|
| Common Voice (ASR)Open ASR Leaderboard | audio-stt | 6.58 % WER | 52 |
| Common Voice (ASR)Open ASR Leaderboard | audio-stt | 4.69 % WER | 52 |
| Common Voice (ASR)Open ASR Leaderboard | audio-stt | 4.51 % WER | 52 |
| Common Voice (ASR)Open ASR Leaderboard | audio-stt | 3.94 % WER | 52 |
| FLEURSOpen ASR Leaderboard | audio-stt | 4.35 % WER | 52 |
| FLEURSOpen ASR Leaderboard | audio-stt | 4.16 % WER | 52 |
| FLEURSOpen ASR Leaderboard | audio-stt | 3.43 % WER | 52 |
| FLEURSOpen ASR Leaderboard | audio-stt | 2.63 % WER | 52 |
| FLEURSOpen ASR Leaderboard | audio-stt | 2.35 % WER | 52 |
Vendor-reported numbers are marked and count with factor 0.7 in rankings. The evidence score says how much a benchmark still tells you today.
Provenance
- Detected
- Aug 25, 2026
- First source
- openasr
- Description
- not yet
- Editorially reviewed
- —