This set of sentences come from the Tatoeba project.
From the approximately 580,000 sentences, I lemmatized every word (giving dictionary forms) within the sentences and deduplicated it according to the lemmatization result. Then, the frequency list from
ruscorpora is used to rank the sentences and remove duplicates. Since it is ordered according to word rank on the frequency table, most sentences should only contain only 1 new word. Audio is obtained using AwesomeTTS.
Update 2021-05-04:
I also added a list of lemmatized words, the target word, and the definition to the card. This should be useful to clarify the meaning of word in question, rather than to be memorized directly.
Update 2021-05-07:
I regenerated the list with a higher quality lemmatization model, which gives a slightly better coverage. I also changed the audio source to Azure, which sounds a lot better.
Please note that this is not perfect; there is a small amount of proper names and compound words not appearing in ruscorpora and might slip in to low-ranked sentences. You can simply delete them or edit them so that they do not use these words. The deck is meant to cover the first 15,000 words, but it does not provide full coverage. Here are some statistics:
Top Coverage
2000 91.3%
5000 87.9%
10000 80.0%
15000 72.5%
Preview:
Please do not attempt to learn a language by solely memorizing vast numbers of cards. This deck is meant to be used as a supplement to learn vocabulary in an easier way than directly memorizing them. However, it is not a replacement for actual immersion and comprehension. I recommend using the roadmap at
https://refold.la/ for a thorough guide to actually learn the language.