Get translation from DeepL via keyboard shortcut
Thread poster: CafeTran Training (X)
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
Sep 8, 2017

While we're anxiously waiting for DeepL's API to be published, you can use some kind of automation to interact with DeepL. Here's a suggestion ...

Send a source segment to DeepL and get your translation inserted in CafeTran's target editor, all via one keyboard shortcut. CafeTran's internal webbrowser is used. In order to save screen real estate, you can decide to hide the source segment editor of DeepL in CafeTran's tabbed pane (or in the floating pane where you placed DeepL).
... See more
While we're anxiously waiting for DeepL's API to be published, you can use some kind of automation to interact with DeepL. Here's a suggestion ...

Send a source segment to DeepL and get your translation inserted in CafeTran's target editor, all via one keyboard shortcut. CafeTran's internal webbrowser is used. In order to save screen real estate, you can decide to hide the source segment editor of DeepL in CafeTran's tabbed pane (or in the floating pane where you placed DeepL).

Mac version:

Demo: https://youtu.be/ChuWgGyPEtI

You'll need Java 8 build 152:


I can confirm the Java 8 issue on Mac with the display of DeepL and a few other web resources is fixed in the coming update of Java 8 too. Early access of Java 8 build 152 (JRE) can be downloaded from here: http://jdk.java.net/8/.


See also: https://cafetran.freshdesk.com/support/discussions/topics/6000050803/page/last#post-6000119776

Windows version:

A demo using AutoHotkey will be published later ...

Linux version:

To be published ...

[Edited at 2017-09-08 07:05 GMT]
Collapse


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
MAC: CMD+A, CMD+C now supported in DeepL's target editor Sep 9, 2017

As of yesterday, CMD+A, CMD+C are supported in DeepL's target editor, when set as a bound editor.

To bind DeepL's target editor as an external editor, choose Edit > Bind external editor. The binding will be indicated by the checkbox "AutoNext":

Screen Shot 2017-09-09 at 09.00.55

So, on Mac the workflow (without using any aids, such as macros) would be:

  • Press ALT+ENTER, to send a source segment to DeepL.
  • Click to position the text cursor in DeepL's suggested translation.
  • Press CMD+A, CMD+C to send the (edited) translation back to CafeTran.


(The macro has the advantage that the second step isn't needed.)


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
Every day an improvement Sep 11, 2017

With today's build, focus will stay in DeepL's target editor. Which makes the sequence of keyboard shortcuts like this:

  • ALT+ENTER, to send the source to DeepL.
  • CTRL+A, CTRL+C (CMD+A, CMD+C on Mac), to get the translation in CafeTran's target editor.


That's three keystrokes to get DeepL's suggestion in your edit box. And then you can start polishing and showing your human supremacy, beating the MT system ...
See more
With today's build, focus will stay in DeepL's target editor. Which makes the sequence of keyboard shortcuts like this:

  • ALT+ENTER, to send the source to DeepL.
  • CTRL+A, CTRL+C (CMD+A, CMD+C on Mac), to get the translation in CafeTran's target editor.


That's three keystrokes to get DeepL's suggestion in your edit box. And then you can start polishing and showing your human supremacy, beating the MT system .

[Edited at 2017-09-12 04:02 GMT]
Collapse


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
My current layout Sep 12, 2017

I have found a nice layout to host the 4 MT engines that I'm currently using for "terminological inspiration".

For this, I've joined Google Translate, Microsoft Translator and Yandex and then docked them to the tabbed pane. DeepL is arranged at the left-hand side.

Google Translate, Microsoft Translator and Yandex are queried automatically upon pressing CMD+OPTION+DOWN (CTRL+ALT+DOWN, on Windows). To query DeepL, I have to press OPTION+ENTER (ALT+ENTER, on Windows) aditi
... See more
I have found a nice layout to host the 4 MT engines that I'm currently using for "terminological inspiration".

For this, I've joined Google Translate, Microsoft Translator and Yandex and then docked them to the tabbed pane. DeepL is arranged at the left-hand side.

Google Translate, Microsoft Translator and Yandex are queried automatically upon pressing CMD+OPTION+DOWN (CTRL+ALT+DOWN, on Windows). To query DeepL, I have to press OPTION+ENTER (ALT+ENTER, on Windows) aditionally.

(On a side note: I'll probably create a simple macro to simulate pressing OPTION+ENTER after my pressing CMD+OPTION+DOWN.)

Note that the translational suggestions from Google Translate, Microsoft Translator and Yandex can be transferred to CafeTran's target editor (the box where you type your translations) via a keyboard shortcut, via dragging over (a part of) the translation or by clicking on the underlined orange link.

Inspirational suggestions from DeepL can currently only be transferred via dragging over them or via CMD+A, CMD+C (provided that you've bound the external editor, and after you've clicked in the suggestion).

Screen Shot 2017-09-12 at 10.13.54
Collapse


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
Simple macro Sep 15, 2017

CafeTran Training wrote:

(On a side note: I'll probably create a simple macro to simulate pressing OPTION+ENTER after my pressing CTRL+OPTION+DOWN.)



I've started using DeepL on my working computer today. I created a simple macro to query DeepL for every new segment:

Screen Shot 2017-09-15 at 09.23.46

I've posted the macro here: https://cafetran.freshdesk.com/support/discussions/topics/6000050803?page=1

For Windows you can create/use a simple AutoHotkey macro that does the same (simulate key press CTRL+ALT+DOWN, wait a little, simulate key press ALT+ENTER).


 
Tom in London
Tom in London
United Kingdom
Local time: 19:19
Member (2008)
Italian to English
Given up on Deepl Sep 15, 2017

I've already given up on Deepl. I decided to try it seriously a week ago, and used it to entirely translate a long academic paper, which I then post-edited.

I discovered that it had totally wasted my time and that I was rewriting the whole translation.

Apart from its lazy mechanical choices and its inability to understand sentence structure, the worst thing about DeepL was that it didn't understand (and never could) that the professor who wrote the paper was developing
... See more
I've already given up on Deepl. I decided to try it seriously a week ago, and used it to entirely translate a long academic paper, which I then post-edited.

I discovered that it had totally wasted my time and that I was rewriting the whole translation.

Apart from its lazy mechanical choices and its inability to understand sentence structure, the worst thing about DeepL was that it didn't understand (and never could) that the professor who wrote the paper was developing her discourse, from the beginning to the end, which was gradually building as the paper progressed. She often mentioned things that implicitly (not explictly) alluded to previous parts of the paper, and it was important that the translation should "sense" this and deal with it.

What I'm learning about MT: it doesn't UNDERSTAND. It simply READS. And misses the whole point. In fact, it doesn't even know there's a point.

To all those CATists out there who think/hope DeepL is going to make your life easier: if your main focus is high-end translations that require the exercise of human intelligence (rather than artificial intelligence) you're going to be disappointed.

[Edited at 2017-09-15 07:43 GMT]
Collapse


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
All true, nevertheless useful Sep 15, 2017

Tom in London wrote:

What I'm learning about MT: it doesn't UNDERSTAND. It simply READS. And misses the whole point. In fact, it doesn't even know there's a point.


That's all true. Fact is, that currently it provides the best translation suggestions for instructive texts (manuals etc.), compared with Google Translate, Microsoft Translator and Yandex.

It often beats them.

Screen Shot 2017-09-15 at 09.43.44

Sorry that DeepL cannot be of more use to you, but very likely you even prefer it that way . It'll keep your brain young.


 
Tom in London
Tom in London
United Kingdom
Local time: 19:19
Member (2008)
Italian to English
Brain Sep 15, 2017

CafeTran Training wrote:


Sorry that DeepL cannot be of more use to you, but very likely you even prefer it that way . It'll keep your brain young.


I fail to see how DeepL could keep my brain young. Or maybe it has already affected yours.


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
Sorry Sep 15, 2017

Tom in London wrote:

CafeTran Training wrote:


Sorry that DeepL cannot be of more use to you, but very likely you even prefer it that way . It'll keep your brain young.


I fail to see how DeepL could keep my brain young. Or maybe it has already affected yours.


Tom, what I meant was this: with DeepL and other MTs one can be tempted to drive/fly on auto pilot. Whereas, when you don't use any of these aids, you'll have to stay sharp with each and every segment, hence, you have to use your brain (and fingers!) more, keeping it young.

[Edited at 2017-09-15 08:18 GMT]


 
John Bishop
John Bishop  Identity Verified
United States
Local time: 14:19
Member (2003)
German to English
Possibly relevant quirk of DeepL's handling of input text Oct 28, 2017

Tom in London wrote: (...) Apart from its lazy mechanical choices and its inability to understand sentence structure, (...)


Regarding the failure to understand sentence structure, I noticed the following during my desultory foolings about with DeepL: if you copy and paste in a large chunk of text, the translator sometimes treats line breaks within the input window as if they were "sentence" breaks (even if there is no punctuation). The output is then pretty poor.

A trick for fixing that is to put the cursor at the beginning of each line, backspace to "connect up" the parts of the sentence, then put the space back in. This turned translations that indeed did not "understand" sentence structure into very good translations of some very complicated (German) sentences.

Anyway, I don't know if your experience might have been affected by this, but I've been very impressed precisely by DeepL's handling of complicated sentence structure once it gets the input right.


 
CafeTran Training (X)
CafeTran Training (X)
Netherlands
Local time: 20:19
TOPIC STARTER
Remove or mask line breaks? Oct 29, 2017

Perhaps it helps to define the line break as a non-translatable? These will then be masked when sending the source segment to DeepL. Or remove the via a macro (AutoHotkey or KeyboardMaestro).

On a side note: the DeepL script isn't necessary anymore, since the support of DeepL has been improved. There's now a keyboard shortcut to insert DeepL's suggestion (when it's of any use to you). See: https://cafetran.freshdesk.com/support/discussions/topics/6000051420

Action to transfer the MT result from a web page to the target segment - see the menu Translate > Transfer MT from web page.


Besides this, you can also mask sensitive data (like company or brand names) from DeepL. All very easy and transparent.

[Edited at 2017-10-29 07:44 GMT]


 
Hans Lenting
Hans Lenting
Netherlands
Member (2006)
German to Dutch
Good news from Cologne! Dec 7, 2017

CafeTran Training wrote:

While we're anxiously waiting for DeepL's API to be published, you can use some kind of automation to interact with DeepL.


Will CafeTran be the first CAT tool to integrate support for DeepL's API?


 
Tom in London
Tom in London
United Kingdom
Local time: 19:19
Member (2008)
Italian to English
Droopl Dec 7, 2017

I've just spent a couple of days experimenting with Deepl to see if if could help or hinder me with a not-particularly-difficult (but not mechanically written) ministerial report on Italy's progress with implementing a particular international agreement.

I did this by copy/pasting the source text, one paragraph or even one sentence at a time, and using Deepl to translate it.

I suppose Deepl may have saved me a bit of time by typing the words for me and guessing most of
... See more
I've just spent a couple of days experimenting with Deepl to see if if could help or hinder me with a not-particularly-difficult (but not mechanically written) ministerial report on Italy's progress with implementing a particular international agreement.

I did this by copy/pasting the source text, one paragraph or even one sentence at a time, and using Deepl to translate it.

I suppose Deepl may have saved me a bit of time by typing the words for me and guessing most of the easy ones, but in every case (i.e. every copy/paste) I had to edit the Deepl translation to make it usable. And it often made really serious errors.
Collapse


 
Hans Lenting
Hans Lenting
Netherlands
Member (2006)
German to Dutch
That's all there is, but it's still much Dec 7, 2017

Tom in London wrote:

I did this by copy/pasting the source text, one paragraph or even one sentence at a time, and using Deepl to translate it.


In my opinion, it's only productive when it's fully integrated in your editor.

I suppose Deepl may have saved me a bit of time by typing the words for me and guessing most of the easy ones, but in every case (i.e. every copy/paste) I had to edit the Deepl translation to make it usable.


Yes, it's a handy typing aid and can give nice jargon clues.


And it often made really serious errors.


Of course, that's where we come in as pros.


 
Hans Lenting
Hans Lenting
Netherlands
Member (2006)
German to Dutch
How DeepL uses the data you send them Dec 23, 2017

The DeepL integration in CafeTran (via its internal Java browser) is still doing a great job. And it's still free. So let us have a look at how DeepL uses the source segments that you upload to them. Here is a statement that you can find when you try to buy an API key:

Screen Shot 2017-12-23 at 09.56.53

BTW: Buying such an API key is currently impossible for freelance translators: you will have to translate more than one million characters per month. Even with CafeTran that will be a challenge.

DeepL writes:

At the moment, we only conclude individual contracts with larger customers. With the volumes mentioned, it makes more sense for both you and us to book API access via a purely online ordering process. However, this will not be available until next year.

I hope you understand that. We will of course notify you when the time comes.


And how about the integration of the new API in CafeTran? The developer writes:

One day in 2018, I may just take a look at their API


So, for the time being, just let us keep enjoying the nice integration via the internal web browser.

[Edited at 2017-12-23 09:21 GMT]


 


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Natalie[Call to this topic]

You can also contact site staff by submitting a support request »

Get translation from DeepL via keyboard shortcut






Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »
Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »