So maybe combine this with DeepMind's recent breakthrough in human voice reproduction, to build an agent that can translate speech in real time and bye bye human translators too ?