I've been working hard to develop an open source CAT tool as part of my teaching work at a German university, the FH Aachen. Here it is:
Although FoundCAT is based on a CAT tool (originally Weblate) it is really designed for learning foreign languages. Learners can upload a text for translating in the crowd, make use of a term base, multiple neural machine translation engines (including DeepL Pro) and of course a translation memory. Learners can also write comments alongside every sentence (segment) they translate while proofreaders can make changes, which are shown as tracked changes. One nice feature is that proofreaders can mark the change with an error category.
I need to do more work to develop FoundCAT so it becomes fun and more intuitive, but I think it's a useful, free tool. Plus FoundCAT is great for those who simply want to play with translating in the crowd. I based the concept on Duolingo Immersion, which is sadly no longer online. The FoundCAT source code is available on GitLab here:
My favorite language combination is German to English (UK), but if you are keen on some more exotic pairs, try German to Arabic.
This is really cool: We have just installed the world's best neural machine translation (NMT) engine, DeepL Pro into FoundCAT. The language combinations are limited to the nine most commonly used, but we're also working on other NMTs so we can offer Arabic, Tamil etc too.
Finally, I've also started to clear up the term base for the language combination German <-> English (UK). All term bases can be imported/exported in various file formats including CSV as well as the standard TBX file.