AI Avatar

Three ways to put a person on screen. Start from a photo and a voice, pick any official HeyGen presenter and your audio, or translate a video you already filmed.

Three tools

Pick the path that matches the footage you have

01 · Photo
Talking Avatar
Upload a portrait and an audio clip. The mouth follows that voice. This is the only path that starts from your own photo.
- Your photo
- Your audio
- Mouth follows the clip
02 · Presenter
HeyGen Digital Twin
You do not upload a face. You pick a name from the full official list, then drive that presenter with audio. Fit and background removal are optional.
- 500 avatars
- Audio required
- Fit
- Background
03 · Translate
HeyGen Video Translate
Start from a video you already filmed, up to 120 seconds. Every official language is listed. Precision favors lip sync. Speed favors turnaround.
- Video ≤ 120s
- 178 languages
- Precision
- Speed

500
Official HeyGen avatars, including the names that keep a trailing space
178
Languages and dialects, including regional variants
120s
Longest source video for translation
MP4
Digital Twin container. It is fixed, not a setting

The frame follows the job

These stills are stand-ins shot on this site, not HeyGen renders. The ratios are the ones the tools actually offer.

16:9

A site or a deck

Digital Twin can set 16:9 when the cut is a walkthrough or an internal update.

3:4

A named presenter

You choose the avatar by name. The default is Abigail Sofa Front.

9:16

A story or a short

The same audio can leave as 9:16 when the cut is vertical.

The form stays small. The rest is in Settings

Media, the model, and aspect sit on the form. Avatar, fit, background, language, and mode open from the gear.

On the form

A photo, a video, or audio
Talking Avatar, Digital Twin, or Video Translate
16:9 or 9:16, when the model has a frame

In Settings

Any of the 500 avatars
Cover or contain
Keep or remove the background
Any of the 178 languages
Precision or Speed

The same face, another language

Video Translate keeps the person in the frame and changes the spoken language. Language and mode are optional. Precision favors lip sync. Speed favors turnaround. The two lines below are a layout, not a finished translation.

We ship on Thursday.

Nous livrons jeudi.

Where this is actually used

A product walkthrough

One named presenter, one audio clip, a 16:9 frame for the site.

A vertical reply

The same audio, 9:16, when the cut is a story or a short.

A dubbed cut

A filmed explanation, translated without reshooting the person.

How a clip leaves this page

01
Choose the tool
Talking Avatar is the default. Digital Twin and Video Translate are in the same menu.
02
Set the file and the options
A photo and audio, or audio plus any avatar, fit, and background, or a video plus any language and mode.
03
Open Create
Generate carries those choices into the studio. HeyGen models are for members.

Questions

Can every model start from my photo?

Talking Avatar can. Digital Twin uses an official HeyGen avatar name. Video Translate starts from a video.

Are any avatars or languages left out?

No. The avatar menu is the full official list, including the names that keep a trailing space. The language menu is the full official list, including regional variants.

Which controls are optional?

On Digital Twin, fit (cover or contain) and background removal are optional. On Video Translate, language and mode are optional. They default to English and Precision.

Who can generate?

HeyGen Digital Twin and Video Translate are for members. This page only prepares the request and opens Create.

Start with the footage you already have

A portrait, a named presenter, or a video with speech. The form above keeps those jobs separate.

Back to the form Image to Video