How to enable and set up machine translation (MT)

Guidelines

  1. Users with the Administrator role must connect a machine translation engine so that linguists can benefit from machine translation matches.

  2. Go to Configuration → Settings → Translation → Machine translation → MT engines, and select an MT engine from the list.

  1. In the General options section, configure the three options there to meet your requirements. The settings you make will be applied globally for all projects (unless overridden by customer or project template settings!):

  • Do not search for an MT match when one of the selected matches exists,

  • Populate target with match,

  • Set MT matches to done.

  1. Below is the list of MT engines, enter the information provided by your machine translation provider to activate the particular machine translation engine.

IMPORTANT!

The connection details a user needs to enter vary from one MT engine to the next!

  • Decide whether or not the use of a particular MT engine is to be a default choice when a project is created. To do so, select the Default setting.

  • Specify Type of account (if applicable).

  • Enter the API Url.

  • Enter the API Key.

  • Enable AI-enhanced TM.

In the case of the Systran MT system, an AI-enhanced TM enables you to improve machine translation matches using the highest Fuzzy Matches coming from the translation memory, up to the 84% of the threshold. Ensure that you deselect the 75-84% fuzzy matches category in Configuration → Settings → Translation → TM → Fuzzy matches, to be able to use this feature. You can view the list of supported languages: List of Supported Languages. If you have a Pro subscription and a dedicated server, the number of available language pairs will depend on the scope of your subscription.

  • Specify when the source text is to be sent off for TM matching by selecting one of the available options in the Send text for matching field:

a) In XTM Workbench → the text is only sent to an MT system when a user enters a segment in XTM Workbench, in "Edit mode". The translation is provided on a segment-to-segment basis.

b) After analysis → the whole text is sent for MT matching during project analysis. When the user enters XTM Workbench, all segments are filled with MT matches.

c) After analysis and in XTM Workbench → the whole text is sent for MT matching during project analysis. The text can be resent to the MT engine on a segment-to-segment basis, as is the case with this option: In XTM Workbench.

  • Decide if you want to make use of the auto inline tag placement feature.

  • Finally, click Synchronize to establish an API connection between XTM Cloud and an MT provider.

  1. Click Save to save the changes.

  2. The next time you create a project, you can choose one of the activated machine translation engines in the Machine translation section to use with your translation and adjust the settings mentioned in item 2 ad hoc.


Setting up Custom MT

Aside from a standard MT engine, you can enable and configure a connection to a custom machine translation engine in XTM Cloud.

  1. Go to Configuration → Settings → Translation → Machine translation → MT engines → Add Custom MT.

  1. A new popup window is displayed. In it, fill all the required fields:

  • Enter a unique Custom MT name for your MT engine.

  • Enter the Username and Password used to log in to the MT engine.

  • Enter the API Base URL.

  • Enter your Authentication endpoint and Translation endpoint.

  • Enter the Check status endpoint.

  • Specify time intervals at which the status is to be checked, in the At intervals of [...] seconds field.

  • Specify your endpoint for downloading translation in the Download translation endpoint field.

  • Choose Custom fields (which are configurable for each MT engine).

  • Optionally, enter a Custom fields prefix (to distinguish custom field metadata sent off to MT from other metadata).

  • Finally, click Add.

The new MT engine is displayed in the Configuration → Settings → Translation → Machine Translation → MT engines section. You can add it when you create a project or in the Project Editor.

IMPORTANT!

Any Custom MT will only work "after analysis", which means that MT matches will be available after a project has been analyzed.


Good to know!

You can also customize the MT setup for specific customers [Customers → (select a customer) → Settings → Machine translation] and a project template [Projects → Templates → (select a project template) → Project details → Machine translation]. The configuration at those levels will of course overwrite global settings.