NVIDIA open-weight batch speech model transcribing 40 languages, including Arabic (rated "Supported"; 6.35% word error on the FLEURS Arabic test). Handles finished recordings up to one hour, speech translation to and from English, and speaker labels. OpenMDW licence. Not live streaming.
- Category
- Voice & speech
- Type
- ASR MODEL
- Country / maker
- NVIDIA
- Pricing
- Price on request
- Documented repository availability
- 27 Oct 2026
- Item last updated
- Not documented yet
- Added to Souq
- 7 Nov 2026
A native speaker reviewed the Arabic description, and the capabilities and languages were checked by hand and approved by the Souq team.
Official links
Official website — huggingface.co (opens in a new tab on huggingface.co)Related tools
© 2026 Souq. Original text and design: all rights reserved. Third-party trademarks and source material belong to their respective owners.