Hume AI is a research company and startup focused on emotionally aware AI, particularly in voice. Its central thesis is that how something is said - tone, pitch, timing - carries as much meaning as the words, and that an AI assistant should both perceive a user’s emotional state from their voice and shape its own vocal responses accordingly. The company is led by Alan Cowen, a researcher in the science of emotion.
On March 25, 2024, Hume announced a $50 million Series B led by EQT Ventures and unveiled its flagship product, the Empathic Voice Interface (EVI). Hume describes EVI as a conversational voice AI “trained on millions of human interactions” that uses a user’s tone of voice for state-of-the-art end-of-turn detection - knowing when a speaker has actually finished - and that learns to generate vocal responses optimized for user satisfaction over time. Hume calls the underlying model an empathic large language model, or eLLM, combining a language model with measurement of vocal expression. A second-generation model, EVI 2, followed in September 2024 with lower latency.
Why business readers should care: Hume represents a distinct branch of the companion-AI wave - one centered on affect and voice rather than text personality. Reading emotion from voice is commercially attractive for customer service, coaching, and assistants, but it also raises sharp questions about consent and manipulation when software is explicitly engineered to detect and respond to a user’s feelings.