I have had audio generated for lessons in Turkish before. But I noticed a few times that it was not complete. Whole sentences were left out. It doesn’t happen in the sentence view. It’s only for the whole text. But with these missing sentences, the audio is not useful for reading along.
This happened to me as well, the entire audio was there, but I could see in the Karaoke that for some reason the audio was jumping from one sentence to another further leaving a lot of sentences in the middle unheard. This was by generating TTS.
Sometimes, also, the audio stopped after few minutes and didn’t move on.
As a workaround, the only thing I could do (on my computer only), was to open the lesson, delete the audio, and generate again a new audio.
Thanks for reporting, we will investigate the issue.