By making speech lighter and easier to integrate, we move closer to AI systems that understand sound with the same confidence they bring to text.' ...