ElevenLabs' new v4 speech model makes AI voices more expressive and consistent

ElevenLabs has released Eleven v4, a speech model designed to improve the consistency and expressiveness of AI-generated voices, including better handling of laughter and whispering. The update includes a Turbo variant optimized for real-time applications with a 150-millisecond latency. By improving performance in long-form content like audiobooks and conversational agents, this release addresses key technical barriers to integrating high-quality, responsive synthetic voices into interactive systems.