AnimeLang
Words read aloud, whatever device you're on
Most phones and computers have no voice for Ukrainian, Czech or Greek at all. The words are now read by a voice on the server instead, so everybody hears them.
Speech used to come from the voices your own device has. On a Mac with the right language installed, that was fine. On most Windows machines, most Android phones and a fresh iPhone it was nothing at all: no Ukrainian voice, no Czech voice, no Greek voice, and a speaker button that did nothing when you pressed it.
That's an acceptable gap in something free. It isn't one in something you pay for.
What changed
Words, examples and whole lines are now read by a voice on the server — one voice per language, the same one every time, so a word doesn't change accent between the drill and the review.
Your device's own voices are still there if you prefer them, and Settings is where you choose. The server's voice is what you get by default, because it's the one that's certain to be there.
Paid for once, not once per person
Anything spoken the same way for everybody — a word, an example sentence, a set phrase — is recorded once and then kept. The next person to tap it hears the recording rather than paying to make it again.
What isn't shared is what nobody else would ever hear: a character's reply to something you wrote, which was made for you and exists for one conversation.