If most of the file is being pre-translated it feels as though there is a time out occurring during the pre-translation.  This may be related to your internet connection... it may be related to some setting on the DeepL server.  I'm not really sure.

Can you share a document with me where this happens for you so i can test it?  Would be interesting to see if this is a general problem or just something occurring for you.  You can send it to pfilkin@sdl.com f this is ok?

I just e-mailed you two files I tested. Thanks for your help!

@Rachel

Unknown said:

Manually inserting the translations is - depending on the file size - painstaking.

For me, having to insert an MT result manually is absolutely no problem. The reason why? However efficient the MT software is, I would never confirm a segment without carefully reviewing and most of the times editing the MT proposal. That's how I don't feel bad starting from MT-(pre-)translated segments and not telling my clients.

BR. Gérard

Gérard LAMBERT said:

I would never confirm a segment without carefully reviewing and most of the times editing the MT proposal.

Gérard LAMBERT

I think you're misunderstanding the workflow here.  I doubt Rachel is automatically confirming either, she just wants to pre-populate the file with NMT matches (draft status) quickly and then she can work offline if she chooses because the document will be fully populated with either a TM match from her memory, or an NMT result where there is nothing in her TM.

@Paul

OK, understood. To tell you the truth, I never willingly pre-translate my files and I do not see them much pre-translated even though pre-translating is supposed to be an automated step of the project creation process.

BR. Gérard

Gérard LAMBERT said:

even though pre-translating is supposed to be an automated step of the project creation process.

Gérard LAMBERT

This is also controlled by how you have defined your project settings...

I was testing a fix in the plugin for something else this morning and purely by accident managed to reproduce the problem you have explained. So, in case this helps you here's what the resolution was to the problem.

You have two locations where you can set the minimum translation match value in your project settings.

Here:

And here:

When you are pre-translating with a TM and also Machine Translation it's important that you take both these settings into consideration.  The first one, under Search is there to set the minimum value at which a result will be returned from your TM at all.  The second one, under Pre-translate Files, is there to set the minimum value at which you wish your file to be pre-translated.

So if you set the Search criteria to 60% and the Pre-translate value to 100% with a TM that could provide matches between 60% and 100% then a pre-translate using MT could result in something like this:

This is because the MT will only be applied where there is no match from your TM based on the Search criteria which was 60%.  In this case the missing segments actually have a TM match at 62%, 63%, 93% and 65%.  All above the 60% I set, but below the 100% pre-translation setting.  Since the option says this:

"Apply MT when no match is found", so it only applies the MT where there is actually no value in the TM at all.  Since the default value in here is 100% failing to change this value too would be easy to miss.  So if I actually wanted to see MT results for all of these I need to tell Studio it's pre-translating to the same value as the Search, so I should raise my minimum match value to 100% in the Search as well.  When I do that the result is this:

If I prefer to see results from my TM for all matches above 60% then I should lower the minimum match value to 60% in the Pre-translate as well.  Then I see this:

I hope that makes sense to you and even more so I hope this resolves the problem you were having?