Is the arcana voice model not working for anyone else whenever you test it with your assistants or is it just me?