Subema on Nostr: So it's audio? Would be surprised they wouldn't have better tech at launch then. ...

So it's audio? Would be surprised they wouldn't have better tech at launch then. Voice generation is pretty far in that, as far as same style/balanced output goes. I mean it's accessible for end user willing to pay pennies. The bleeding edge models with scriptable intonation/speed/volume/idontevenknow are circulating in the papers and huggingface demos. So, if I were to bet, voice quality wouldn't be the reason for fail