This is pretty dang cool but I guess I was hoping to see some very clever use of vision and maybe waveform analysis to give the model "eyes and ears" to edit audio and video. V2