Open ASR Leaderboard · English short-form
Open ASR Leaderboard · English short-form is run by Hugging Face hf-audio. It has ranked 65 models, of which the catalogue holds 26, scoring from 4.0614 to 6.0414 on avg (Average WER %).
measured by Hugging Face hf-audio · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 65 | Speech-to-text (Scribe v2) | avg (Average WER %) | 4.0614 |
| 2ndof 65 | vfast | avg (Average WER %) | 4.1314 |
| 3rdof 65 | resonant-1 | avg (Average WER %) | 4.15 |
| 4thof 65 | resonant-1-flash | avg (Average WER %) | 4.1629 |
| 5thof 65 | scribe_v1 | avg (Average WER %) | 4.2086 |
| 6thof 65 | azure-speech-06-2026 | avg (Average WER %) | 4.2371 |
| 7thof 65 | Hojo-ASR-V1 | avg (Average WER %) | 4.29 |
| 8thof 65 | Qwen3-ASR-1.7B | avg (Average WER %) | 4.4257 |
| 9thof 65 | higgs-audio-v3-stt | avg (Average WER %) | 4.45 |
| 10thof 65 | Universal-3.5 Pro | avg (Average WER %) | 4.4614 |
| 11thof 65 | pulse | avg (Average WER %) | 4.54 |
| 12thof 65 | canary-qwen-2.5b | avg (Average WER %) | 4.55 |
| 13thof 65 | ARK-ASR-3B | avg (Average WER %) | 4.61 |
| 14thof 65 | granite-speech-4.1-2b | avg (Average WER %) | 4.68 |
| 15thof 65 | solaria-3 | avg (Average WER %) | 4.7 |
| 16thof 65 | ARK-ASR-0.6B | avg (Average WER %) | 4.72 |
| 17thof 65 | MOSS-Transcribe-Diarize | avg (Average WER %) | 4.74 |
| 18thof 65 | granite-speech-4.1-2b-nar | avg (Average WER %) | 4.74 |
| 19thof 65 | MOSS-Transcribe-preview-2B | avg (Average WER %) | 4.76 |
| 20thof 65 | parakeet-tdt-0.6b-v2 | avg (Average WER %) | 4.82 |
| 21stof 65 | cohere-transcribe-03-2026 | avg (Average WER %) | 4.84 |
| 22ndof 65 | Parakeet TDT 0.6B v3 | avg (Average WER %) | 4.96 |
| 23rdof 65 | canary-1b-flash | avg (Average WER %) | 5.03 |
| 24thof 65 | Phi-4-multimodal-instruct | avg (Average WER %) | 5.1 |
| 25thof 65 | granite-4.0-1b-speech | avg (Average WER %) | 5.13 |
| 47thof 65 | Whisper Large v3 | avg (Average WER %) | 6.0414 |
Read from the board on 2026-08-25