It took me perhaps an entire minute before I realized something was strange about the voiceover and that it was synthesized, which, in my view, is quite impressive. Once I realized it was generated, though, I became much more attuned to the pitch-shifting artifacts, slight cadence irregularities, and cultural misses (like emphasizing the second syllable of "recall" in the title of the movie Total Recall).

Still, a few more years of progress on neural nets (can GANs be applied to vocal cadence and intonation?) and I won't envy the position of mid-tier voiceover artists.