Hello Dima Dima

From what I know is not possible to sell the license to another user. A solution would be to purchase a 12 months subscription instead of a perpetual license. You can contact our Customer Experience team here for more details: www.sdltrados.com/.../chat.html or https://sdl.com/chat / https://www.sdltrados.com/store/

Thank you,

Sorin

Dima Dima Sorin Perta

Just to add to what Sorin mentioned.  The 12-month subscription is probably a good choice, or even the Starter version for a one off project.  But if you do purchase a perpetual licence you can sell it on if you wish.  However, this is something you will have to organise with whoever you sell it to as this will involve you handing over your account to that user too.  We do not provide any tools for managing such a marketplace.

Thank you Paul! So it´s in principle possible to transfer the license. That sounds good. ... 

Maybe one other question. I have seen your video about Google Translate V3 MT. I wanted to ask, ... according to your expereance, is it worth to train your own translation model with Auto ML. Does the Michine Translation delivers better results, when it works with custom translation models?! Thats the main reason, why I would use SDL Trados Studio, because it has the Google Translate V3 plugin.

Dima Dima

Dima Dima said:

is it worth to train your own translation model with Auto ML. Does the Michine Translation delivers better results, when it works with custom translation models?

Well... I'd say yes if you have the right sort of data to feed it.  I don't think this is a trivial thing and if you are not a translator, and are only doing this one book with a CAT then what are you going to use to train the model with?  I'd also ask that you consider the costs as the use of trained models in Google is not cheap and for a one-off project like this is it really worth it.. especially a book.  What kind of content is it?

Dima Dima

Dima Dima said:

is it worth to train your own translation model with Auto ML. Does the Michine Translation delivers better results, when it works with custom translation models?

Well... I'd say yes if you have the right sort of data to feed it.  I don't think this is a trivial thing and if you are not a translator, and are only doing this one book with a CAT then what are you going to use to train the model with?  I'd also ask that you consider the costs as the use of trained models in Google is not cheap and for a one-off project like this is it really worth it.. especially a book.  What kind of content is it?

I am in complete agreement with you Paul. It's not worth it for a single project. But I have the possibility to use Tanslation Memory from the translation of a book with a similar theme. I translated this book some time ago in the traditional way, with the help of Machine Translation, of course. But the main work was done by myself. Now I would like to try an experiment and translate a second book, but now with the help of the new possibilities offered by Google Translate V3. It is a book about palmistry. More precisely, about palm reading according to Vedic knowledge. So it's astrology, where you don't work with birth charts, but with the palms. The content of this book is a mixture of "technical" explanations and philosophical worldview. Therein lies the difficulty for Machine Translation. I would actually only do this experiment out of sporting interest. As I said, I am not a translator, but I find the subject of Machine Translation very interesting. I am clear that machine translation will never work like the mind of a human, because that is not possible in principle and will never be possible. So for me it's a kind of interesting gimmick.

Dima Dima

Perhaps you should take a look at OPUS-CAT:

https://appstore.sdl.com/language/app/opus-cat-nmt/1160/

This is a free machine translation solution you can run yourself on your local computer.  You can also train the engines with the translation memory you have, so perhaps a sensible place to start rather than pay for Google AutoML which is expensive and complex to set up.

That looks very interesting. I don´t know the quality of the basic MT of OPUS-CAT, but it is worth it to try it out. Well google´s MT engine is pretty stong, but as you already said, also very expensive to train. Thank you Paul! 

I do have one final clarifying question. Because it´s not finaly clear, how it will be handled by SDL.  If I buy a SDL Trados Studio license and register it in my name, then this license will be managed by SDL under my name, with my email address. If I now no longer need the license after the completion of my project and sell - the buyer of the license then simply takes over my account at SDL or can the license key, through SDL support, be transferred to his own account?

Translated with www.DeepL.com/Translator (free version)

Dima Dima

Dima Dima said:

the buyer of the license then simply takes over my account at SDL

Correct.

Thank you Paul!