New York-based startup Hume has introduced its latest AI model, EVI 3, designed to enhance customer support systems, health coaching, storytelling, and virtual companionship. This conversational AI model allows users to create custom voices by interacting with it, offering naturalness, expressiveness, and empathy. EVI 3 is tailored for businesses, developers, and creators, providing advanced customization, faster responses, and improved emotional understanding.
Developers and businesses can leverage EVI 3 for various purposes, such as customer service, content creation, and in-app interactions. The model enables users to specify personality traits, vocal qualities, emotional tone, and conversation topics, resulting in engaging and humanlike interactions. While lacking voice cloning capabilities, Hume prioritizes flexible voice customization and plans to incorporate voice cloning features in the future.
Internal benchmarks reveal that users prefer EVI 3 over competitors like OpenAI’s GPT-4o, Google’s Gemini model family, and other AI models. With features like prosody generation, interruptibility, in-conversation voice customizability, and API-ready architecture, EVI 3 offers robust multilingual support, low latency, and unlimited custom voices. The model’s pricing is usage-based, with enterprise discounts available for large deployments, making it accessible to a wide range of users.
Hume, founded in 2021 by Alan Cowen, aims to enhance human emotional nuance in AI interactions. The company’s extensive dataset, trained on participants worldwide, captures speech, text, vocal bursts, and facial expressions to improve emotional intelligence in AI interfaces. With a history of emotive AI voice models like EVI 2 and Octave, Hume continues to innovate in the field of voice AI, empowering developers and creators to explore new possibilities in artificial intelligence.
