“Christel Heydemann predicts that achieving seamless, low-latency, real-time voice translation at scale on mobile networks will take a few years, but not decades.”