Hume has introduced its third-generation speech-language model, EVI 3, which brings more expressiveness, realism, and emotional understanding to voice AI experiences. EVI 3 can be fully personalized with any voice and personality created by a user prompt, allowing for instant generation of new voices and personalities. This is achieved through Hume's latest research on speech-language models, which developed methods to capture the full range of human voices and speaking styles in one model. EVI 3 has been evaluated against other leading voice-to-voice AI models, including GPT-4o, Gemini, and Sesame, with favorable results in areas such as emotional tone modulation, expressiveness, and naturalness. The model is capable of delivering voice responses in under 300ms on state-of-the-art hardware and is currently available through a live demo and iOS app, with API access planned for release in the coming weeks.