Occitan translation and speech-to-text models oc

1 model into Occitan, 1 from Occitan, 12 speech-to-text.

Which open models work for Occitan?

Windy Translate lists 1 open translation model into Occitan and 1 from Occitan. All of them can be downloaded from Hugging Face and run on your own hardware. By licence: 2 CC-BY-4.0 (terms and attribution text: licensing). None of them has a screening score yet: FLORES-200, the test set used here, does not cover these pairs, so test them on your own text before you rely on them.

With English. English to Occitan: translate-tc-big-en-cat_oci_spa (not yet scored). Occitan to English: translate-tc-big-cat_oci_spa-en (not yet scored). To translate between Occitan and a language with no direct model, you can go through English with these two.

Speech-to-text. Whisper's tokenizer lists Occitan as one of its languages, so all 12 multilingual Whisper-based models here can transcribe it: see the speech models. Accuracy differs a lot between languages and model sizes, and this site has not measured speech accuracy.

Each model page has a short snippet to download and run the model yourself, its licence and attribution text, and its score where there is one. Pick the best scored model for your direction, then try it on a few sentences of your own text: a screening score is a starting point, not a guarantee.

Into Occitan

From Occitan

Scores: chrF++ on FLORES-200 dev, 48 sentences, measured 2026-08-04. Method.