Real-time voice intelligence for every application

Remarkably human-like personalities powered by our empathic voice-to-voice model.

Try Demo

Kora

Stella

Dacher

Whimsy

Aura

Ito

Dacher

Whimsy

Kora

Stella

Dacher

Whimsy

Aura

Expressivity

Femininity

Speed

Pitch

Accent

Extroversion

Raspiness

Formality

Rhythm

Enunciation

Expressivity

Femininity

Speed

Hume LLM

Web Search

Tool Use

External LLM

Custom LLM

TTS Injection

Hume LLM

Web Search

Tool Use

External LLM

Custom LLM

TTS Injection

Hume LLM

NPC

Assistant

Coach

Agent

Tutor

Clinician

App UI

NPC

Assistant

Coach

Twilio

Typescript

Python

React

API

Twilio

Typescript

Python

React

Trusted By

Flagship Model: EVI 2

Our latest voice-to-voice model converses rapidly and fluently with users, understands users' tone of voice, and generates the right tone of voice. Capable of emulating a wide range of personalities, accents, and speaking styles, it can be tailored to each application and user.

Start Building

Our model capabilities

Multimodal emotional intelligence

EVI 2 merges language and voice into a single model trained specifically for emotional intelligence, enabling it to emphasize the right words, laugh or sigh at appropriate times, and much more, guided by prompting to suit your use case.

Learn More

Voice customization without the risks

Create synthetic voices unique to any app or user, without voice cloning. Our novel approach lets you modulate EVI 2’s voice along dimensions like timbre, pitch, nasality, perceived gender, and more, then control its tone and speaking style with your prompt.

Learn More

Support for any LLM and tool

Let EVI 2 generate all the language or integrate any external LLM or tool. EVI 2 will seamlessly incorporate any output into the conversation without sacrificing on expressiveness, personality, speaking style, or instruction-following capabilities.

Learn More

Explore our capabilities

Compelling personalities (Aura) with EVI 2

"Hey Aura..."

Empathically expressive speech with EVI 2

"I’m launching something I'm excited about…"

Compelling personalities (Whimsy) with EVI 2

"Hey Whimsy..."

Rapping on command with EVI 2

"Can you freestyle rap about yourself?"

Prompting rate of speech with EVI 2

"Can you speak faster from now on?"

Nonverbal vocalizations with EVI 2

"Could you laugh maniacally for us?"

Inventing new vocal expressions with EVI 2

"Now can you make a sound of joy and enthusiasm?"

Emergent multilingual capabilities with EVI 2

"Can you speak Spanish?"

Compelling personalities (Stella) with EVI 2

"Hey Stella..."

00/00

Empathic Voice Interface Pricing

EVI 2 (Beta)

$0.072 / min

Improved transcription, language modeling, and speech generation by a single voice-to-voice model with better empathic responses
Extensive voice customization with 7 base voices, adjustable parameters, and language-based voice prompting
Faster performance with latency of 500ms - 800ms
Support for English with more languages coming soon

Details

EVI 1

$0.102 / min (Legacy)

Transcription, language modeling, and text-to-speech coordinated across models to generate empathic responses
Customizable voice options with 3 base voices
Latency range of 900ms - 2000ms
English language support only

Details