Hume AI is an AI-powered data annotation platform that enables businesses to create high-quality training data for machine learning models.

Hume provides the AI toolkit to measure, understand, and improve how technology affects human emotion. Use our models to measure human expressive behavior in images, videos, audio, and text and build applications that better predict users’ intents, preferences, and outcomes. Coming soon: better LLMs, aligned with real signs of user satisfaction and well-being.

Aegis Ventures

Wisdom Ventures

Comcast Ventures

Northwell Holdings

Union Square Ventures

LG Technology Ventures

Evan Sharp

Hume AI

Our emotional expressions are the building blocks of empathy and social connection. But for a long time, scientists have sorted them into only a handful of categories, greatly underestimating their complexity. 

At Hume AI, we’re building a new science of expression that measures the full spectrum of nuanced expressive behaviors people form every day. We’ve run experiments around the world and discovered that humans worldwide share more than 25 dimensions of meaning in our facial and vocal expressions. 

We’re translating these insights into the most accurate machine learning models ever built for understanding human expression. And we’re helping developers around the world build technology that supports human emotional well-being.

Want to learn more? Watch the video below!

Hume's expression measurement models are derived from semantic space theory (SST), a revolutionary computational approach to mapping out emotional behaviors in a dimensional space pioneered by our CEO & Chief Scientist Alan Cowen.

Before SST, researchers relied on outdated classification methods like the Basic 6 emotions and Facial Action Coding System (FACS), which is limited by its reliance on the human eye's capacity to discern features, often leading to subjective interpretations. By introducing computational methods to emotion science, we’ve discovered that human emotion is high-dimensional, blended, and categorizable.

In our latest blog by Jeff Brooks, discover why textbooks are being rewritten around semantic space theory: https://lnkd.in/ejMmiwWD

Hume AI shows that our feelings are way more complicated than just being happy, sad, or scared. Using Semantic Space Theory (SST), we've learned that there are over 20 distinct emotional dimensions beyond the basic six categories. This data-driven approach offers a nuanced understanding and computational modeling of emotions across cultures.

https://lnkd.in/e4vGkDpT

"What we cherish" is an innovative project by Gehl that uses Hume's models to explore the emotional landscape of our cities. It shows how AI can translate soft, qualitative data into deep visual insights — and turn feelings into art.  

Gehl, a leading architecture and research consultancy that focuses on putting life at the center of urban design, asked people to record stories on one simple question: what do you love about your neighborhood?

But communication is about so much more than just what we say — it's how we say it. That's why Gehl used Hume's AI models to decode the emotional expressions in speech prosody. We need to interpret the rhythm, stress, and intonation in speech for a richer understanding of community stories. Learn more about our models, available through a public API: 
https://lnkd.in/gXAQGiWF

Imagine seeing the feelings of your community. That's what Gehl did next: they visualized our expression measurements, turning each emotional tone into a soft shape with a musical note. This creates a unique audiovisual symphony.

This project celebrates the emotional richness of our communities by helping us understand, visualise, and cherish our connections. Explore more:
https://lnkd.in/gfb7FsWf

AI:HUME

Hume AI

Analysis Not Available