Language reqests

[PJ-Finlay]

Thread for language requests.

Rules:

• Top level comments only names of languages
• Don't duplicate top level comments (Use Ctrl-F to search this page before commenting)
• Upvote the language models you want added so that the devs can gauge demand for different languages

Resources for training your own models

• Toolkit for training/converting LibreTranslate compatible language models 🚂
• Argos Train
• Custom trained models are available for $1000/each.

[PJ-Finlay]

Slovak

[Suvis-1]

Tamil too please

[waqargpgc]

please add these languages somalia language and Hmong language and Burmese (Myanmar) language @PJ-Finlay

[waqargpgc]

any update somalia and Burmese (Myanmar) languages @PJ-Finlay

[Rc-Ali]

Please add Hmong Burmese and Soomaali @PJ-Finlay 😁

[mojrs10]

Croatian pls

[PJ-Finlay]

Czech

[PJ-Finlay]

#87

[PJ-Finlay]

Done

[nanmu-hm]

KHMER

[PJ-Finlay]

Indonesian

[PJ-Finlay]

I think this is Indonesian? https://www.youtube.com/watch?v=W_EzM2WrHB4

[cmaree]

That YouTube is from an Indonesian YouTuber, but her example uses English <> Russian. It seems we still need the Bahasa Indonesia translation. Thank you!

[PJ-Finlay]

Greek

[PJ-Finlay]

#95

[PJ-Finlay]

Done

[hendursaga]

Is this just Modern Greek?

[PJ-Finlay]

Is this just Modern Greek?

I think the data for the current model is mostly modern Greek. I made a separate request for "Ancient Greek" language:
#91

[telmobss1]

Esperanto

[binrebin]

Malayalam

[idclang]

Malayalam

[aandrich3]

Esperanto can be very useful as an intermediary language.

[PJ-Finlay]

Done

[nikitaindik]

Dutch

[argosopentech]

Done

[nikitaindik]

Wow! Thanks a lot! ❤️

[mhmdaji]

arabic-german

[vk0xOrg]

Ukrainian

[argosopentech]

Done

[vk0xOrg]

thank you !

[mmahmoudian]

Finnish 🇫🇮

[PJ-Finlay]

Done!

[mmahmoudian]

Persian (also known by its endonym "Farsi")

[telmobss1]

it would be nice to have a latinized form under the word in Persian script

[argosopentech]

TIL "endonym" - A name used by a group or category of people to refer to themselves or their language, as opposed to a name given to them by other groups. (Wiktionary)

[PJ-Finlay]

Done

[ghost]

Hungarian (magyar)

[argosopentech]

Done!

[ghost]

Thank you @argosopentech!

Are you able to advise when Hungarian will be showing at https://libretranslate.com/?

Thank you once again!

[PJ-Finlay]

Hungarian should be available on new LibreTranslate deploys already. I think https://libretranslate.com redeploys on every commit to the main GitHub repo so probably not that long.

[ghost]

Hungarian is now showing at https://libretranslate.com/. Köszönöm szépen!

[rohaan2614]

Urdu

[PJ-Finlay]

Done! I added a Opus-MT Urdu model.

[hussain-mustafa990]

@PJ-Finlay How do I added Urdu model to my installation, i am unable to have it being listed by using the command "argospm search" but nothing pops up for urdu.

[BastardoSenpai]

Swedish

[PJ-Finlay]

Done!

[italotoffolo]

Latin

[hendursaga]

Support for Neo-Latin would be great!

[pclahoud]

Bump for latin

[PJ-Finlay]

https://community.libretranslate.com/t/train-model-for-latin/998

[FelixFrizzy]

Latin would be great indeed

[italotoffolo]

Hebrew

[GreenLunar]

Awesome! Thank you!

I was hoping I'll do it first, but I'm still waiting for ArchLinux to
fix their package in AUR.

Once I finally get the packages installed, is it possible to add to the
current Hebrew package?

Did anyone test that package yet?

[alexzlotnikov]

Any updates?

[PJ-Finlay]

Hebrew is supported

[tabassum-alphadna]

Bengali

[MoollaMore]

How can the translation for this language be improved? from english

[argosopentech]

German-Finnish

[argosopentech]

Hindi-French

[ansonhevu]

It would be very useful Portuguese (Brazil) to Spanish and also Spanish to Portuguese (Brazil) because in South America there are lots of spanish countries and Brasil near them so it would simplify talking between these countries. Using english as pivot intermediate can result in errors... so this is why it is important.

[ansonhevu]

I perceived in my tests that the english pivot intermediate introduce translation errors. To be considered an useful and professional software Argos needs direct language pairs and needs to eliminate the english pivot intermediate. Without direct languages pairs Argos is a toy, not a tool. I compared Portuguese (Brasil) to Spanish between Argos and msty + ollama + llama model and the translation results are much better with llama and the problem is the english pivot intermediate in Argos which introduce vocabulary and grammar errors. Argos has a great potential but developers should focus in release direct language pairs...

[JoanEliot]

Norwegian (Bokmål)

[r4z137]

Mongolean

[r4z137]

i know not many people need that - and that would make a difference if you support it ;)

[Tsengel-de]

hi there just created a MR/PR to add our language model in argos-translate 😄
argosopentech/argospm-index#8
and here are the models MN->EN, EN->MN (https://huggingface.co/Tsengel-mn/argos-mongolian) ... if he do not want merge it we can create our own fork and develop further, 😅 . if u have any questions, just let me know ...

[sbotirov]

Uzbek language, please 🙏

Thanks in advance!

[ljubomirm]

Please add the Slovenian language.

[chxrlie]

Welsh (Cymraeg) 🏴󠁧󠁢󠁷󠁬󠁳󠁿

[Scoty]

Bulgarian

[wayubi]

Pashto

[wayubi]

Dari

[jaggedsoft]

Macedonian

[m-rey]

Toki Pona

[art-over-group]

Turkmen