Windy TranslateCatalan translation and speech-to-text models ca
9 models into Catalan, 9 from Catalan, 14 speech-to-text. Best scored into Catalan: Portuguese to Catalan (chrF++ 55.8). Best scored from Catalan: Catalan to French (chrF++ 54.1).
Which open models work for Catalan?
Windy Translate lists 9 open translation models into Catalan and 9 from Catalan. All of them can be downloaded from Hugging Face and run on your own hardware. By licence: 16 Apache-2.0 and 2 CC-BY-4.0 (terms and attribution text: licensing). 16 of them have a screening score on FLORES-200. The best scored into Catalan is translate-pt-ca (chrF++ 55.8, Good). The best scored from Catalan is translate-ca-fr (chrF++ 54.1, Good).
With English. English to Catalan: translate-en-ca (chrF++ 54.6, Good), the best of 2 direct models. Catalan to English: translate-ca-en (chrF++ 53.7, Good), the best of 2 direct models. To translate between Catalan and a language with no direct model, you can go through English with these two.
By band, the scored Catalan models are 10 Good, 5 Usable and 1 Limited (Excellent is chrF++ 60 or more, Good 45 to 60, Usable 30 to 45, Limited 15 to 30). A score is a 48-sentence screening of news and encyclopedia text, not a measure of accuracy, and a difference of a point or two is noise. How we measured.
Speech-to-text. 2 speech-to-text models made for Catalan: listen-windy-lingua-ca and listen-windy-lingua-ca-ct2. Also, Whisper's tokenizer lists Catalan 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 Catalan
| Model | chrF++ (FLORES-200 dev) | Licence |
|---|---|---|
| Portuguese to Catalan best scored translate-pt-ca | 55.8 Good | Apache-2.0 |
| French to Catalan translate-fr-ca | 55.4 Good | Apache-2.0 |
| English to Catalan translate-en-ca | 54.6 Good | Apache-2.0 |
| Spanish to Catalan translate-es-ca | 53.5 Good | Apache-2.0 |
| Italian to Catalan translate-it-ca | 49.8 Good | Apache-2.0 |
| German to Catalan translate-de-ca | 43.9 Usable | Apache-2.0 |
| Dutch to Catalan translate-nl-ca | 42.9 Usable | Apache-2.0 |
| Ukrainian to Catalan translate-uk-ca | 35.8 Usable | Apache-2.0 |
| English to Catalan, Occitan and Spanish translate-tc-big-en-cat_oci_spa | not in FLORES-200 | CC-BY-4.0 |
From Catalan
| Model | chrF++ (FLORES-200 dev) | Licence |
|---|---|---|
| Catalan to French best scored translate-ca-fr | 54.1 Good | Apache-2.0 |
| Catalan to English translate-ca-en | 53.7 Good | Apache-2.0 |
| Catalan to Portuguese translate-ca-pt | 52.3 Good | Apache-2.0 |
| Catalan to Spanish translate-ca-es | 51.0 Good | Apache-2.0 |
| Catalan to Italian translate-ca-it | 47.1 Good | Apache-2.0 |
| Catalan to Dutch translate-ca-nl | 43.8 Usable | Apache-2.0 |
| Catalan to German translate-ca-de | 43.5 Usable | Apache-2.0 |
| Catalan to Ukrainian translate-ca-uk | 28.9 Limited | Apache-2.0 |
| Catalan, Occitan and Spanish to English translate-tc-big-cat_oci_spa-en | not in FLORES-200 | CC-BY-4.0 |
Catalan speech-to-text
| Model | Formats | Licence |
|---|---|---|
| listen-windy-lingua-ca | safetensors | Apache-2.0 |
| listen-windy-lingua-ca-ct2 | ct2-int8 | Apache-2.0 |
Scores: chrF++ on FLORES-200 dev, 48 sentences, measured 2026-08-04. Method.