Identified why the tongue model can't position R sounds
Welcome back
Found the model learns relative shape changes but strips the absolute anatomical data needed for tongue placement.
Derived from this session's token and cost data. Not shown on the feed.
Comments