Microsoft

Hub: Building a custom system using a dictionary only

Now available: Customize neural and speech translation with the new Custom Translator

Custom Translator allows you to customize Microsoft Translator’s neural machine translation when translating text using the Microsoft Translator Text API v3, and speech translation using unified Speech services.

When used with Speech, you can also use the Custom Speech portal to customize speech transcription to better understand the user’s voice, subject matter, or work better in noisy environments, and the Custom Voice portal to customize the text to speech voice output.

Learn more


You may build a custom system on Translator Hub without using parallel documents by using a dictionary only, no other files needed. 

This will work for projects in language pairs with a fully supported Microsoft translation system behind it (see below), not for any other language pairs.

A dictionary is an Excel file containing the language code as column header in row 1, and the values of phrases you want translated the exact specified way in the subsequent rows. You specify "Do not translate" by entering the same phrase twice. You may enter as many languages as you like in the same file, and the columns can be in any order. 




Then you select this file in the “Dictionary” tab.




If there are less than 2000 sentences available, there is no point choosing anything in the “Training” tab. Unselect any files in the Training tab that may have been automatically selected.
You are now ready to train and deploy a system. You will not get a test report. The training in this case will only extract your dictionary, and deployment will publish the dictionary as part of the Microsoft Translator service.

Language pairs with a fully supported Microsoft translation system behind it

English (en) from and to any of the following:

af Afrikaans
ar Arabic
bg Bulgarian
bn Bengali
bs-Latn Bosnian – Latin
ca Catalan
cs Czech
cy Welsh
da Danish
de German
el Greek
es Spanish
et Estonian
fa Persian
fi Finnish
fil Filipino
fj Fijian
fr French
he Hebrew
hi Hindi
hr Croatian
ht Haitian Creole
hu Hungarian
id Indonesian
it Italian
ja Japanese
ko Korean
lt Lithuanian
lv Latvian
mg Malagasy
ms Malaysian
mt Maltese
mww Hmong (White)
nl Dutch
no Norwegian
otq Querétaro Otomi
pl Polish
pt Portuguese
ro Romanian
ru Russian
sk Slovak
sl Slovenian
sm Samoan
sr-Cyrl Serbian (Cyrillic)
sr-Latn Serbian (Latin)
sv Swedish
sw Kiswahili
th Thai
tlh Klingon
tlh-Qaak Klingon (pIqaD)
to Tongan
tr Turkish
ty Tahitian
uk Ukrainian
ur Urdu
vi Vietnamese
yua Yucatec Maya
zh-HanS Simplified Chinese

Japanese (ja) from and to Korean (ko)


Edited on October 25, 2018

Feedback and Knowledge Base