Omnilingual ASR - Advancing automatic speech recognition for 1,600+ languages
Meta's Omnilingual ASR is an open-source (Apache 2.0) speech recognition model supporting 1,600+ languages. It uses an LLM-based architecture that can be extended to new languages with just a few in-context examples, without retraining.
Replies
Flowtica Scribe
Hi everyone!
Meta FAIR just open-sourced the ASR model that supports 1,600 languages. (Yes you read that right, 1,600.)
It's released under an Apache 2.0 license and covers almost every low-resource language you can think of. No matter what voice app you're building, this could be a powerful supplement to your main ASR model, letting you reach a much wider audience.
Language connects the world, but so many smaller languages are disappearing. Meta also turned the dataset into an interactive language exploration map where you can listen to the languages. It's a great way to experience the world's linguistic diversity.
Swytchcode
1600 is a big number! Is it the best out there (as of today)?