Sesame Unveils Its Expressive Conversational Voice Demo

Sesame's February 2025 voice preview explores how conversation history can shape the way an assistant speaks. Its technical article describes a Conversational Speech Model that uses text and prior audio when generating speech. That context helps address a basic ambiguity: the same sentence can be spoken many ways, and the appropriate tone or rhythm depends on the exchange around it. The preview's companions were optimized for friendliness and expressivity. The accompanying vision of an all-day wearable companion was a longer-term ambition, not a hardware release.