Amazon's new text-to-speech model shows "emergent abilities"

[dupe]

More discussion here: https://news.ycombinator.com/item?id=39373814