1) It's far from being that easy. Even if you take a DVD/Bluray source, the quality is all over the place depending on what studio made it or other factors and with streamable ones, even studios that have above average video quality like Knights Visual still only use ok aac quality for the audio.
2) -mp3 is super old(31 years since its creation), the only reason it's still used today is because everything that plays audio support it pretty much.
-aac is also very old(25 years since its creation). The quality is ok and it's probably the most widely used one for high compression audio these days.
-opus is better than both of those(10 years since its creation). It's not that widely spread yet, mostly on streaming platforms wanting to compress their audio to save on bandwidth cost.
Simplified, the older the codec, the worst the quality will be for the same bitrate(file size), but if you give them enough bitrate, they become all the same quality pretty much so which is best depends on that bitrate. 320kbps mp3 will be better than 128kbps aac for example but 192kbps mp3 will be worse than 128kbps aac.
Also, the less transformed it is the better since every time you compress with a lossy compression(most of them are when released to the public), quality will degrade.
I was curious about that and tried looking at the code to find it but I couldn't by just searching for the word "translate" and follow the code a bit.
Edit: Deducing from the answer to
this question, seems like it's using its own knowledge of language to translate and not an external service.