| Pages in topic: < [1 2 3] > | Poll: Do you make use of machine translation options inside your CAT tool(s)? Thread poster: ProZ.com Staff
|
|---|
Stepan Konev 
Russian Federation
Local time: 08:07
English to Russian
What regards the subject question, yes I do use OPUS CAT, an offline free NMT engine compatible with my CAT tools. It retrieves suggestions from my TM in my language based on my previous translations.
| | | | Alex Lichanow
Germany
Local time: 07:07
Member (2020)
English to German
+ ...
| TM suggestions | Feb 3, 2023 |
Stepan Konev wrote:
What regards the subject question, yes I do use OPUS CAT, an offline free NMT engine compatible with my CAT tools. It retrieves suggestions from my TM in my language based on my previous translations.
Trados Studio has been doing that since the 2019 version (I think) through the "wrench" suggestions. It's a hit and a miss in many cases, but it doesn't requre any additional plugins, so I am happy to just stick to that. Also, MT in itself is a hit and a miss in most cases, anyway. Honestly, I mostly only see garbage from most MT engines that are not called DeepL, and I consider my time and income to be way too valuable to waste it on editing machine garbage.
EDIT: Maybe I should also mention that I work in what must be the most common language pair for any new and eXcItInG technologies popping up on a monthly basis, so I don't even want to imagine what machines do to more "exotic" language pairs.
[Edited at 2023-02-03 07:33 GMT]
| | | | Nikolay Novitskiy
Russian Federation
Local time: 10:07
Member (2018)
English to Russian
I translate video games with stories, and I enjoy doing it creatively.
MT has a very dull and "uncreative" language, so my answer is NO. Under no circumstances.
| | | | Dan Lucas
United Kingdom
Local time: 06:07
Member (2014)
Japanese to English
Stepan Konev wrote:
What regards the subject question, yes I do use OPUS CAT, an offline free NMT engine compatible with my CAT tools. It retrieves suggestions from my TM in my language based on my previous translations.
Stepan, how difficult has the process of setup, configuration, and maintenance been? I looked at OPUS a few weeks ago and it looked fairly onerous, at least for JA-EN. I probably wouldn't get permission to use it even as an offline tool, but I am interested...
Dan
| | |
|
|
|
| Touché ... Fair Enough ... I am talking about translating more general-content texts | Feb 3, 2023 |
Ice Scream wrote: I don't use MT full stop. I think constant exposure to computer-generated language would dull my linguistic instincts and take away half the fun of the creative process. Like using TMs provided by an agency. Justin Peterson wrote:
At this point, in 2023, I see not using a good MT tech inside a good CAT as the equivalent of not using a computer, but sticking to a typewriter Many of us don't translate simple, repetitive texts, in which case neither MT nor CAT will be an advantage.
| | | | Lieven Malaise
Belgium
Local time: 07:07
Member (2020)
French to Dutch
+ ...
| Some thoughts | Feb 3, 2023 |
Maciej Bogucki wrote:
Sometimes I feel like the only person on this planet who thinks that the output of all MT is still unreliable garbage, and yes, that includes DeepL.
It's not garbage in my language pairs, but it is indeed (and luckily) unreliable. Imagine it wouldn't be ...
Maciej Bogucki wrote:
And believe me - this machine-suggested writing is never going to be as good as doing it yourself from scratch.
Interesting to see how little confidence translators have in their own skills. It's entirely up to the translator what the result of MT editing will be.
Maciej Bogucki wrote:
Nowadays I think I do about as much edition as I do translation, and at this point I can recognise MT or "MT-assisted" translation within 15 minutes, because it always produces the same kinds of glaring red flags. I'm sorry, but anyone who thinks this is something that produces good translation that can be used within a professional capacity is deluded. In fact it's downright horrifying to me how often I have to deal with these kinds of hackjobs these days, where the translator lets a machine do the majority of the work for him and then calls it a day after applying the smallest modicum of effort to edit it (if even that). It's simply a terrible practice that drags this entire profession into the mud.
So you have to deal with very bad translators. Don't blame MT for that.
| | | | Lieven Malaise
Belgium
Local time: 07:07
Member (2020)
French to Dutch
+ ...
Stepan Konev wrote:
What regards the subject question, yes I do use OPUS CAT, an offline free NMT engine compatible with my CAT tools. It retrieves suggestions from my TM in my language based on my previous translations.
So you are basically recycling your own translations. That's exactly the opposite of what I consider to be a real advantage of a (good) MT auto-suggest function. It offers you angles you might not think about immediately, and thus broadens your translation horizon.
| | | |
Stepan Konev wrote:
Funnily, even without constant exposure, some fellow translators don't seem to understand the original question: do you use MT within your CAT tool? They read it like 'What do you think about MT?'
You don't just expect me to roll over, do you?!
yes I do use OPUS CAT, an offline free NMT engine compatible with my CAT tools. It retrieves suggestions from my TM in my language based on my previous translations.
Now that is interesting. I didn't know they could do that. Learn from the master
Could possibly be worth experimenting with. Is it straightforward to set up?
| | |
|
|
|
Stepan Konev
Russian Federation
Local time: 08:07
English to Russian
| QTranslate+OPUS | Feb 3, 2023 |
Lieven Malaise wrote:
So you are basically recycling your own translations. That's exactly the opposite of what I consider to be a real advantage of a (good) MT auto-suggest function. It offers you angles you might not think about immediately, and thus broadens your translation horizon. You can use DeepL for that. In my setup, I always have two MT apps: OPUS for offline suggestions and Qtranslate that inserts DeepL/Google/Yandex suggestions.
| | | | Stepan Konev
Russian Federation
Local time: 08:07
English to Russian
| OPUS installation | Feb 3, 2023 |
Dan Lucas wrote:
Stepan, how difficult has the process of setup, configuration, and maintenance been? I looked at OPUS a few weeks ago and it looked fairly onerous, at least for JA-EN. I probably wouldn't get permission to use it even as an offline tool, but I am interested Hi Dan. I don't think it is difficult. You need to install the app itself and a plugin for Trados. Then you follow the instructions to fine-tune the engine using your TM (tmx). It doesn't require special skills. Here is how to: https://helsinki-nlp.github.io/OPUS-CAT/install
| | | | Stepan Konev
Russian Federation
Local time: 08:07
English to Russian
Alex Lichanow wrote:
Trados Studio has been doing that since the 2019 version (I think) through the "wrench" suggestions. Right, but the fuzzy repair feature in Trados can't create new content based on what you have in your TM. It can only slightly modify what already exists. On the other hand, OPUS builds whole sentences.
Honestly, I mostly only see garbage from most MT engines We often see what we want to see. There are thousand reports of successful use of DeepL or other MT engines. I see people who now admit using tools or MT "for typing assistance only" and who completely denied it a while ago. And I often see people who say that a tool is overestimated because it has too many features they don't need. It is not the tool that is overestimated. It is them who "underuse" the tool. It's a matter of habit, and our habits and beliefs are always hard to change. It's ok, it's by design.
| | | | Samuel Murray
Netherlands
Local time: 07:07
Member (2006)
English to Afrikaans
+ ...
No. I create an extraction of the segments, translate it in the machine, align it and create a secondary TM with the machine translation in it, and penalize the TM. Despite my CAT tools having built-in MT capabilities. I sometimes also make edits to the MT TM before or during translation, to improve the quality of the suggestions (something which isn't possible with live MT).
| | |
|
|
|
| @Samuel: edits to the MT TM | Feb 3, 2023 |
Samuel Murray wrote:
No. I create an extraction of the segments, translate it in the machine, align it and create a secondary TM with the machine translation in it, and penalize the TM. Despite my CAT tools having built-in MT capabilities. I sometimes also make edits to the MT TM before or during translation, to improve the quality of the suggestions (something which isn't possible with live MT).
I guess that you are not referring to injecting your preferred terms into MT suggestions?
And: Is there no way to skip the alignment step?
| | | | | Pretranslation is OFF TOPIC, but... | Feb 3, 2023 |
Hans Lenting wrote:
And: Is there no way to skip the alignment step?
eTranslation made on 15.04.2020 from a source TXT file the translated TXT file(s)
http://www.condak.cz/nove/2020-04/15/cs/02.html
More than one year it is possible to get the output file in TMX or XLIFF format.
---
eTranslation can also be used to translate files in the same format. Here is the result of my 60 minutes of hard work:
https://www.condak.cz/nove/2023-01/28/cs/00.html
I translated the Czech EPUB into 31 languages. You can POST-EDIT them. If you know how.
Milan
| | | | Baran Keki
Türkiye
Local time: 08:07
Member
English to Turkish
| Not related to the topic being discussed but... | Feb 3, 2023 |
Ice Scream wrote:
Like using TMs provided by an agency.
Having to use the TM provided by an agency is like wearing someone else's dirty socks, most of the time. And having your rates reduced based on the matches coming from that TM... Are you supposed to or expected to or do you feel duty-bound to correct/re-translate a 100% match if it's complete and utter rubbish? Or do you simply confirm it and proceed to the next one because you're paid only 10% of your full rate for those matches? And if you do that, would you shift the blame on to the previous/other translator if the client complained of the quality of translation? Or do you take the rap for that?
Speaking of agency TMs/CAT tools, why the hell do agencies have your name displayed in full at the bottom of their CAT tool, revealing your identity to the reviewer (meaning the other translator, who's likely to know who you are through Proz)? Whatever happened to confidentiality, anonymity, 'right to be forgotten' and all that shit? After having signed and scanned all those Confidentiality rubbish for them, they don't extend you the same effing courtesy?
| | | | | Pages in topic: < [1 2 3] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: Do you make use of machine translation options inside your CAT tool(s)? | TM-Town | Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
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 » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
