Hume AI EVI 2

EVI 2 is Hume AI's voice-to-voice foundation model: it merges language and voice, trained for emotional intelligence to enable remarkably human-like, real-time conversations.

Hume AI's Empathic Voice Interface 2 (EVI 2) is the foundational voice-to-voice model engineered for high emotional intelligence. It processes a user's tone and adapts its response, generating any required tone with subsecond (500ms-800ms) latency for fluid, real-time dialogue. The model offers extensive customization: developers can adjust base voices on continuous scales (gender, pitch) and maintain a wide range of dynamic personalities and multilingual capabilities. EVI 2 is available via API, focusing on emotionally-attuned interactions while strictly avoiding voice cloning risks.

https://www.hume.ai