Introducing HeyGen Voice

The world's #1 voice model. Now in HeyGen and via API

Hi sundoelcoy,

The world's best voice model is live in your account!

HeyGen Voice, our in-house voice model, debuted at #1 on Artificial Analysis' voice leaderboard, ahead of ElevenLabs v4, ElevenLabs v4 Turbo, and Alibaba's Qwen Audio. It's also the first voice model built for video.

Text to Speech Leaderboard (Controlled Voice)

Elo

HeyGen Voice HeyGen

1,201

Qwen-Audio-3.1-TTS-Plus Alibaba

1,182

Eleven v4 Turbo ElevenLabs

1,166

Eleven v4 ElevenLabs

1,162

Sonic 3.6 Cartesia

1,142

Every model voices the same 8 speakers, 4 US and 4 UK English. October 2026. View the leaderboard

It performs your script in your voice, with your pacing and expression, in the acoustics of whatever scene you put it in.

Learn more

What's new in your voice settings

Adaptive mode

Keeps your voice and fits the delivery and acoustics to the scene image.

Exact mode

Keeps your voice and your recording conditions together, so new lines sound like the same session, no matter how long your video is.

Expressiveness Boost

One slider from your baseline delivery to full energy, without losing what makes the voice yours.

Build with the API

HeyGen Voice is available through the HeyGen API at $30 per million characters. No packages, no minimums.

View the API docs

Start creating

Getting started takes 60 seconds. Record a short sample on your phone, set HeyGen Voice as your default in Settings, and you'll hear yourself in a new scene within minutes.

Instant Voice

Free
Made from 60 seconds of audio

Capture your voice for free

Professional Voice

Paid add-on
Made from up to 2 hours of audio

Create a Professional Voice