I heard 'mangiamo pesce sono te', which has no sense at all, but I simply couldn't hear 'stanotte' at all. Why can't they record these texts by natives???
yes, they use TTS ( text to speach). if you want to know more, there are discussions in general section about should it be TTS, recordings by professionals, natives, or even by users themselves. I prefer current option.
I agree that TTS is probably the best bet for Duolingo at this point in time. Eventually they probably should record native speakers in a controlled studio to get consistency and a more natural audio feature.
In my experience, its much harder to understand native speakers until you have a good mind for the context of a conversation. Whenever I learn languages, I always find the most difficulty with comprehension.