06 nov Underdog shows off a 32MB speech model in new demo

Posted at 10:02h in Tech-En by

Despite its small size, Underdog’s new model can generate both speech and music.