The Emotional Intelligence Lab for Voice AI

Providing the open source models, datasets, and evaluation APIs to embed emotional intelligence into your voice models.

Our decades of research in multimodal emotional intelligence span

50+

Languages

48+

Emotions

600+

Voice Descriptors

Human Feedback API

Running Human Studies Has Never Been Easier

Use our science-backed survey templates designed to get the most important, scientifically grounded preference feedback for your model.

Blog Post

Why Are Human Evals Needed?

Automated metrics only tell part of the story. Learn why human evaluation is essential for building voice AI that truly connects with people.

Coming Soon

RESTful API

Simple endpoints to create and manage evaluation studies programmatically.

High-Quality Ratings

Draw from a worldwide pool of reliable, vetted participants to get unbiased human feedback on your model.

Fast Turnaround

Receive human preference data in hours, not weeks.

Data

Train on the Best Voice AI Data with Proven Results

Access a library of curated speech datasets for the hardest and most important voice AI tasks.

Dataset Composition

Gaming & Esports
Discussion & Talk Shows
Lifestyle, Food & Vlogs
Entertainment & Performing Arts
Sports
Religion & Spirituality
Education & Academia
Business & Finance
Politics & Government
Health & Wellness
Blog Post

Building Better Voice AI with Curated Data

How high-quality, annotated speech datasets can dramatically improve model performance across languages and domains.

Coming Soon

Conversational Audio

Turn-taking, interruptions, and multi-speaker dynamics.

Emotional Reproduction

Fine-grained annotations across 48 core emotions.

Multilingual Audio

Native speaker recordings across 50+ languages.

Voice Realism

Prosody, intonation, pacing, and expressive range.

Domain-Specific

Industry-tailored datasets for healthcare, finance, and more.

Task-Specific

Workflow-oriented data for scheduling, support, and onboarding.

Models

Equip Your Team with the Latest, Cutting-Edge Research

We believe the best voice AI is built through collective effort. Keep up to date as we open-source more SOTA models and resources.

T

TADA

Open Source

Our open-source LLM TTS system that streams text and audio together, reducing hallucinations and latency.

O

Octave

Closed Source

Our closed-source LLM TTS system with voice design, voice modulation, voice cloning, voice conversion, and more.

E

EVI

Closed Source

Our closed-source LLM Speech-to-Speech system with interruptibility, back channeling, expressive instruction following, external LLM compatibility, and more.

Get Started with Hume Today

Build, train, and evaluate your voice AI models with us. Reach out to get started.

Stay in the loop

Get the latest on empathic AI research, product updates, and company news.

Join the community

Connect with other developers, share projects, and get help from the team.

Join our Discord