Omnilingual ASR

Omnilingual ASR logo - Product Hunt launch

Advancing automatic speech recognition for 1,600+ languages

#Open Source #Artificial Intelligence #Audio
SUMMARY

Meta's Omnilingual ASR is an open-source speech recognition model supporting over 1,600 languages, utilizing an LLM-based architecture to extend to new languages with minimal examples and no retraining. Released under the Apache 2.0 license, it covers numerous low-resource languages, enhancing voice applications' reach. Additionally, Meta offers an interactive language exploration map to experience linguistic diversity.