👋 Welcome to EMOVA! We are a team focusing on fully open-sourced omni-modal foundational models with visual, textual, and speech capabilities. EMOVA (EMotionally Omni-present Voice Assistant) is a novel Omni-modal Large Language Model with end-to-end speech capabilities while maintaining state-of-the-art vision-language performance. We wish to promote the development of omni-modal human interactions with intelligent models!