integreat-cms
integreat-cms copied to clipboard
Support automatic translations via Intento
Motivation
Integreat and associated translations should take less time and process to maintain. Also it leads to lower costs of the whole project.
Proposed Solution
DeepL API can be connected to the cms as well as the Intento API. Therefore we should find a way how to connect our Translation Memory to the cms to reuse the professionall translations from past and get rid of MemoQ as far as this is possible.
Alternatives
If TM-Connecting is too difficult at the start we could begin with the glossary (term database) and the enhance it with TM on a later date.
Additional Context
~Duplicate of #1104~
Ok sorry, didn't notice the Intento part... this isn't supported yet.
@ulliholtgrave @timoludwig prio high can be removed. don't focus that in the next 2 month...
What I also take from the description and have not found in the open issues is that we want to connect our TM from MemoQ, do you want to open an extra issue for that?
What I also take from the description and have not found in the open issues is that we want to connect our TM from MemoQ, do you want to open an extra issue for that?
Indeed, could you open a separate issue for this? Ideally with a detailed description how you intend to use the TM.
@dkehne Did we already decide to implement Google Translate or are we still in the process of deciding? I think we can already create another issue regarding the "additional attribute that defines the preferred automatic translation provider", right?
As this ticket is not yet well defined this ticket will be moved into 24Q1 for now.
@MizukiTemma Google Translate Prio 1, Microsoft Translate Prio 2. Based on DeepL-quality-Report: https://www.deepl.com/en/quality.html
Altough in general scorings Google seems to be a bit better.
☝️Keep #2586 in mind before/when implementing this issue.
It's really hard to tell at the moment if we go with the Cloud Translate API Basic (~costs around 20$/1m chars) or Cloud Translate API Advanced (~costs around 80$/1m chars). The advanced API can be trained with our data from the .tmx-Files.
Hopefully you can implement the API without forcing us to make this decision too early. I guess we need somehow both options to compare the results...So maybe go for the Basic-API first and then mabye add the API-Advanced as a second API in the test cms only?!
Hopefully you can implement the API without forcing us to make this decision too early. I guess we need somehow both options to compare the results...So maybe go for the Basic-API first and then mabye add the API-Advanced as a second API in the test cms only?!
The implementation was changed so both basic (v2) and advanced (v3) versions can be used, by specifying the variable GOOGLE_TRANSLATE_VERSION