@neauoire I mean it's trying to translate a 3D spatial language that combines shapes and movement of two hands, facial expressions and body language, all in parallel to a 1D stream of symbols. That's bound to be seriously difficult!