๐ The ๐ค Open ASR Leaderboard evaluates open-source and proprietary speech recognition models on English and multiple European languages.
We report the Average WER (โฌ๏ธ lower the better) and RTFx (โฌ๏ธ higher the better). Models are ranked based on their Average WER, from lowest to highest. Check the '๐ค About' tab to understand how models are evaluated.
To promote transparancy and reproducibility, evaluation scripts can be found on GitHub. Further trends and observations can be found in our paper.
If you want results for a model (or dataset) not listed here, you can directly open a pull request (PR) on GitHub!
Select columns to display:
- Toggling datasets will change the Average WER.
- Rank ฮ shows how the ranking changes compared to the average over the default datasets (AMI, Earnings22, Gigaspeech, LS Clean, LS Other, SPGISpeech, Tedlium, Voxpopuli).
- Private data (scripted) and Private data (conversational) columns show aggregated WER on private datasets from Appen Inc. and DataoceanAI (see the 'Private data' tab for more info).
Last updated on 5 May 2026