alt
Hacker News
usef-
•
yesterday at 10:35 PM
•
0 replies
•
view on HN
I think they're claiming it's achieved by text models, not voice models, fwiw.