Hume AI Review 2026

Empathic voice-AI platform (EVI) that models emotional expression in speech, producing conversational AI with prosody, laughs, and sighs — plus expression-measurement APIs.

Hume AI is a New York-based startup founded by former Google DeepMind researcher Alan Cowen, focused on 'empathic AI' — modeling the emotional expression carried by voice, face, and language. Its flagship product is the Empathic Voice Interface (EVI), a speech-in / speech-out API where the model doesn't just transcribe and generate text; it also perceives and produces prosody (pitch, rhythm, pauses, laughter, sighs) that reflects emotion. The result is a conversational AI that sounds more human — used in mental-health support (with clinical partners), customer-support voice bots, coaching apps, tutors, and interactive characters in games. Hume also sells an Expression Measurement API for analyzing facial, vocal, and language expression from audio/video files. Pricing for EVI is usage-based: a Free tier with monthly credits and slow ramp-up rates, Starter tiers in the low single-dollar-per-hour of conversation range, and higher tiers ($7+/hour of conversation and up) for premium models, faster response times, and higher rate limits, with Enterprise contracts for volume. Because it competes with ElevenLabs, OpenAI Realtime, Cartesia, and Retell in the increasingly crowded 'realtime voice' space, Hume differentiates hard on emotional expressiveness and its research heritage in affective computing. For product teams building anything conversational — coaching, wellness, education, gaming NPCs — Hume is often the pick when the interaction quality matters more than raw voice latency, though its models have improved on speed as well through 2025 and 2026.