Applying a second vocal model doesnt seem to work