Translation:They would have looked at the animals.
Here's a funny thing: in the fast audio, the phrase is clearly includes "elles_auraient" (liaison between the final s and the verb) and "des animaux" (clear "d" in "des), but in the slow audio, there is no audible s in "elles" (so it could be singular "elle aurait") and the direct object is audibly "les animaux" (clear "l" in "les").
No access to slow audio from here, although i am not surprised liaisons are skipped, the tts must be breaking that up word by word. To be fair, if a person were to break down a sentence word by word, they would just as likely skip the liaison as well as it sounds very awkward to attach it to either the preceding or following word. I'd probably try to linger between the two with a long zzzzzz sound, nothing quite natural either.
In the fast one i hear "les". It's not a given that we are listening to the same engine though.
All in all, it's probably better not to expect too much from the text to speech. It's really good as these things go, but it's still a very tough technological nut to crack and is bound to be imperfect for a while still. I also wouldn't be surprised if the French tts here were performing a bit less convincingly than English, Spanish or German - those languages have always seemed to be higher priority on Duolingo, with understandable reasons.
Good luck with your learning, you seem to have more than one pot on the fire there!
But the problem, as mentioned earlier, is that the liaison is lost when you use the slow version (which I did in trying to hear it more clearly.) So the slow version doesn't just slow down the audio; it pronounces each word individually, which ironically makes some words harder to understand. Frustrating, but I try to remind myself that listening to actual conversation would be even harder.