Hi! I have loved this service - it has helped me amazingly. Now, however, after 1 million words of reading, it’s becoming less and less useful. It’s not easy to get pdf books in Greek and the different videos available (under headings such as history, news, sports or whatever) are actually more trouble than they are worth. The problem is that speech-to-text is still very weak. AI doesn’t know how to say “I don’t know” so that you can put any combination of letters and it will come up with a guess. (If the AI translation is the only one available that’s a pretty sure sign it’s not a word at all) So I spend A LOT of time editing the texts. To cap it all, I’ve just realised that even if you correct the text (takes time), it will still include the nonsense/inexistant words into your LingQs created list, so you have to go there too to remove them. Goodness, I was so determined to make sure my LingQs didn’t include nonsense words, false translations - that project failed
Thank you for your message, and thank you as well for the thoughtful feedback.
It is great to hear LingQ has helped you so much, especially to the point of reaching 1 million words read. I also understand your frustration. If the transcript quality is poor and you then have to spend a lot of time correcting text and cleaning up incorrect LingQs, that takes away from the learning experience.
Transcript quality can vary a lot depending on the source, especially with auto-generated captions or speech-to-text imports. In cases like that, the best approach is usually to edit the transcript before creating many LingQs, since incorrect spellings in the source text can lead to incorrect LingQs being created as well.
If the content comes from video sources, manually added captions or a proper transcript usually work much better than auto-generated subtitles. If you have access to the text separately, importing clean text instead of relying on speech-to-text can also help.
I appreciate you pointing this out. Feedback like this is very useful, and I will pass it along to our team, especially your comments about inaccurate transcript words being added to the LingQs created list even after corrections.
Google Play books has ebooks in Greek, although I can only guess how many, as there is no direct way to search for books in a specific language. But using calibre you can turn it into a pdf.
Thalia.de has 2130 ebooks in Greek. Although I am not familiar with their drm, as I’ve never bought an ebook there, it seems that it can be similarly bypassed like with Google Play books. The same applies to Hugendubel.de.
Regarding the messed up ai transcriptions: You could try downloading the videos and use whisper to transcribe it, using a better language modul. In addition, you might let ChatGPT or another ai do the corrections. It might not yield perfect results, but could save you some work.
Hi Obsttorte, and thank you very much. Yes, I have realised that my problem is not being able to use Calibre - as I don’t have an ebook reader and you need a serial number to be able to run the DRM remover. My mistake was falling back on all those videos that we see on our pages under various titles such as history, politics and whatever. In Greek, they are all terribly unreliable, and, as I suggested, more trouble than they are worth. I have ebooks that I can’t upload - no doubt because of DRM. Greek learners are SO HIGHLY indebted to Rafarafa ! For the moment, he is saving me! Thank you for your response. Best wishes, Orla
HI Zoran, could I possibly send you an ebook or two to see if they are genuinely protected by DRM? They don’t look like they are - but I haven’t managed to upload any of the PDF ebooks I have… thank you for your help. Best wishes, Orla
I don’t think this is the case. I have a kindle, but you don’t need one to read amazon ebooks. The ebooks I buy on Google Play don’t require one either. They can be read with Adobe Digital Editions.
The serial number is not tied to some reader, but to your account, me thinks.
I’ve read millions of words in Greek with Lingq. I find Greek hard but this method works for me.
I buy interesting books on https://www.psichogios.gr/el
I mix reading them and ministories or simple AI texts (which aren’t great but the Greek is ok).
I convert them to text using Calibre or ultimate epubor. There may be a step with adobe digital editions.
It may be useful to take out contents pages and publisher text at the beginning.
You can end up with large amounts of interesting Greek text to read.
@bbbblinq You should quote me correctly, so
Obsttorte:
[…] n […]
I am not quiet sure what part of what I have written you are refering to, to be honest.
Adobe Digital Editions is required for books bought on Google Play Books in order to download them. The resulting file can either be a PDF or an EPUB. The first case can be problematic, as some PDF books are just a series of pictures, not suitable for beeing imported into LingQ. However, it is often the case one can choose.
Importing the file into Calibre removes the DRM using the DeDRM plugin. You have to specify the passkey in the settings. The same accounts for books from Amazon, although there it only works for books published before a specific date. (There seems to be an updated version for newer books, but the process is more complex and I haven’t tested it yet).
There’s the rub! And I have bought books through Amazon and on the screen they open with Kindle, but that doesn’t give you a serial number (I did my homework: perhaps in the past they did, now they don’t). So, without a reader device, you are blocked. I guess I’ll just have to hunt down a friend who owns one…
It appears that the new way of bypassing the DRM may not require the serial number. Make sure you have download the bought ebooks using the Kindle Reader App. Download DeDRM 10.0.28. The zip contains a file named MSIXKFXArchiverMobi1_18632.exe. Executing it will create a folder called archifed_kfx, which will contain the encrypted books that you can then open in calibre. I just tested it with a book released 12/25 and it worked like a charme.