Hello,
I have trados studio 2019.
I have been using automated translation for a while, but recently the quality has impaired.
I'm using english (american) to french.
It occurs both with sdl and with google.
I tried to find an answer elsewhere but didn't succeed.
As it happens with both google and sdl, I suppose I've messed up somewhere. But where ?
The translation has become a word by word translation.
Thank you if you can help !
Kris
The translation has become a word by word translation.
Do you mean the sentences are segmented by word? Does this happen with every file you translate? Or what exactly do you mean by this... can you share a screenshot of anything to show it?
Paul Filkin | RWS
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
Hello,
No the segmentation is fine, but the translation is not made by sentences, with a full meaning anymore. The automated translation is made by word and the result is mostly without any meaning. Gramar and syntax have become incorrect.
Thank you.
can you try copy pasting the segment source into Google Translate online and see if it's any different? I don't think there is much we can do to change this.
You could also try using SDL Machine Translation which is also available to you in Studio and the new NMT is on a par with the best of them.
Paul Filkin | RWS
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
@Paul Thank you for your answer. It is better on google translate. On google translate it is as it used to be before.
@Paul Thank you for your answer. It is better on google translate. On google translate it is as it used to be before.
I found a solution, which is to convert my pdf document into Word with another OCR converter before adding it to studio. The google api then works much better. But I suppose the issue isn't really fixed if any other person gets the same problem without a OCR software.
I think you just made it clear to me what the problem is... wish I'd thought of this before. You may be hiding the formatting tags that are probably in the file due to it being a PDF. If you press Ctrl+Shift+H it's likely you have a lot of tags in there. These may be "playing" with the results by interfering with the way the text is being handled.
If you can verify this I think we should look at a "sending text only" option in the settings to avoid problems like this when the tagsare not needed but for a CAT tool they are there.
Paul Filkin | RWS
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
Hey you're right ! Loads of tags. Ok I'll try this option. Thank you.
Ok I'll try this option.
There isn't one yet!! I said "I think WE should look at" this for a future update.
For now your best bet is to clean the PDF before translating. This is usually the best approach anyway. Another converter is a good approach if you have one, or use something like TransTools which will clean up the PDF really nicely.
Paul Filkin | RWS
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
Translate
