Fact · Speech recognition
1600+languages
Meta's Omnilingual ASR is an open-source speech recognition model family for more than 1,600 languages.
Checked against its source on October 17, 2026
In the source's own words
“Omnilingual ASR: Open-Source Multilingual Speech Recognition for 1600+ Languages”
omniASR-LLM-7B model card (Omnilingual ASR), Meta FAIR, Hugging Face, 2025, Page title.
Open the original- What it measures
- The model card's title states the language count.
- What it does not show
- Coverage is not accuracy. A language count says nothing about error rate in each language, so test your own language and audio before you rely on it.