Skip to content

How-to · 19 engines · up to 10 min · free plan

Make a photo talk — one picture, one voice, four steps

Pick a portrait, give it a voice and choose an engine: that is all it takes to make a picture talk. This guide walks through each step, which photos work, how to get the audio and which engine to pick for speech, singing or two people.

  • Upload or record the voice
  • Paintings and cartoons work too
  • Two people in one photo
  • Up to 10 minutes of speech

Updated by the

Made with Kenerate

Quick answer

Updated

How do you make a photo talk with AI?

To make a photo talk with AI you need 2 files: a front-facing portrait and a voice track. In Kenerate's Avatar Talk you upload the picture, upload or record the speech, pick one of 19 talking-photo engines and generate; the engine moves the lips, face and head in time with every word, for up to 10 minutes.

Key facts

Steps
4
Talking-photo engines
19
Longest speech
10 min
Photo
Up to 20 MB
Audio
MP3, WAV, M4A and more · up to 50 MB · or the mic
Free plan
Yes

Real results

Pictures that talk: five kinds of photo

A family portrait, an old painting, a cartoon, a singer and a presenter. Real Wan 3 clips with generated voices; in Avatar Talk the words come from your own audio.

AI-generated people and characters. Open a clip to copy its prompts.

How to

How to make a picture talk, step by step

Time needed
Steps
4 steps
  1. Step 1

    Choose the photo

    A sharp face looking at the camera, mouth closed, nothing covering it; any image up to 20 MB. Paintings, cartoons and 3D characters work too.

  2. Step 2

    Give it a voice

    Upload a voice file (MP3, WAV, M4A and more, up to 50 MB), record it on the mic, or write the words in Voice Studio and pick the take from History.

  3. Step 3

    Pick an engine

    Kenerate Avatar is preselected and takes up to 10 minutes; OmniHuman 1.5 suits singing; InfiniteTalk Duo gives two people in one photo a voice each.

  4. Step 4

    Generate and save

    Add an optional direction such as "warm and friendly, gentle nods", generate, then download or share the video from your history.

Kenerate AI Avatar Talk mode card set to Talking photo with the Kenerate Avatar engine

Open Avatar Talk, pick Talking photo, script to video or lip sync, and choose the engine.

1 / 5 · Choose the mode and engine

Watch the tour · 2:23

Talking photos, script to video, engines, any language, presenters, lip sync & chat

Chapters

Checked 6 October 2026

Best way to make your picture talk

Start from what you have. Each row names the engine or mode that fits it best, from Avatar Talk's engine list.

Best way to make your picture talk — checked 6 October 2026
What you haveBest for itWhy
One portrait and a long scriptKenerate AvatarTakes up to 10 minutes of speech, in 720p or 1080p
A photo and a songOmniHuman 1.5Animates singing and full-body movement, up to 60 seconds
Two people in one photoInfiniteTalk Duo or LongCat 1.5 DuoEach person gets their own audio track
A face where every word must look rightSync Lipsync 3Frame-accurate mouth shapes, up to 5 minutes
No usable photoHeyGen Presenters500 stock presenters, 16:9 or 9:16, up to 4K
A video, not a photoLip sync mode (VEED Lipsync 2)Keeps the original performance and redraws only the mouth

All rows run in Avatar Talk; switch engines in the engine picker.

Sources: Kenerate Avatar Talk · Kenerate Voice Studio · OmniHuman-1 project page

Engines

Which engine makes your photo talk best

Seven of the 19 talking-photo engines in Avatar Talk, matched to the kind of picture and audio you have.

  • Kenerate Avatar title card — a long-form presenter video still of a fictional adult presenter at a desk in a modern studio
    Default

    Kenerate

    Kenerate Avatar

    Input
    Photo + up to 10 min of speech
    Output
    720p or 1080p

    Long talks and the sharpest everyday result

    Open Kenerate Avatar
  • OmniHuman 1.5 title card — an expressive full-body fictional adult performer singing on stage with dramatic lights

    ByteDance

    OmniHuman 1.5

    Input
    Photo + up to 60 s of audio
    Output
    Engine decides

    Singing photos and full-body performance

    Open OmniHuman 1.5
  • 2 speakers

    WaveSpeed

    InfiniteTalk Duo

    Input
    Photo of two + one track each
    Output
    480p or 720p

    Two people in one picture

    Open InfiniteTalk Duo
  • Sync Lipsync 3 title card — a fictional adult speaker's lips synced to a waveform that stretches to match a longer audio track

    Sync Labs

    Sync Lipsync 3

    Input
    Face photo + up to 5 min of speech
    Output
    Engine decides

    Mouth shapes that match every word

    Open Sync Lipsync 3
  • Skywork

    SkyReels V3 Avatar

    Input
    Photo + up to 15 s of audio
    Output
    480p or 720p

    Short clips with a described emotion

    Open SkyReels V3 Avatar
  • Lightricks

    LTX 2.3 Lipsync

    Input
    5–20 s of speech; the photo is optional
    Output
    480p, 720p or 1080p

    Sharp short clips for social posts

    Open LTX 2.3 Lipsync
All models (7)
  • HeyGen Presenters title card — a lineup of ready-made fictional adult presenters standing in a sleek broadcast studio

    HeyGen

    HeyGen Presenters

    Input
    Audio + one of 500 stock presenters
    Output
    720p, 1080p or 4K

    No photo of your own

    Open HeyGen Presenters

Speech limits and resolutions are from Avatar Talk's engine list. The gallery clips were made with Wan 3 image to video.

Tips

Five-second checks before you generate

Most odd-looking results come from the photo or the audio, not the engine.

  • Crop to head and shoulders; tiny faces in group shots sync poorly.

  • Pick a photo with the mouth closed and teeth hidden.

  • Remove background music from the voice track first.

  • Old or blurry photos: restore or upscale them before animating.

  • Keep the first test short, then run the full script.

  • Only animate photos of yourself, your own characters or people who agreed.

Example uses

How creators use Make a photo Talk

Illustrative examples of typical workflows, not customer reviews.

  • A woman looking through an old photo album at a kitchen table

    Family historian

    A family historian could make a grandparent's portrait tell a short story recorded by a relative, for a family reunion slideshow.

  • An educator with a tablet beside a classical portrait in a gallery

    Museum educator

    A museum educator could make a painted portrait introduce itself in a short gallery-guide clip for visitors' phones.

  • An illustrator's desk with cartoon character sketches on the wall

    Illustrator

    An illustrator could make a cartoon character from a book say its catchphrase in a short teaser video.

FAQ

Make a photo talk — questions

How do I make a picture talk for free?

Sign up and open Avatar Talk: it works on the free plan with the sign-up credits. Upload the photo, add or record the voice, keep Kenerate Avatar as the engine and generate. Short clips use less, so test with a sentence or two before you run a long script.

Link to this answer

Can I make a photo talk with my own voice?

Yes. Choose Record audio and speak into the microphone, or upload a recording you already have. Avatar Talk accepts common audio files such as MP3, WAV, M4A, AAC, OGG and FLAC up to 50 MB.

Link to this answer

Can I type the words instead of recording them?

Yes, in two steps. Write the script in Voice Studio and generate the voice there, then in Avatar Talk pick that take from History as the audio. The link to Voice Studio sits right under the audio box.

Link to this answer

Can I make an old family photo talk?

Yes, if the face is clear and looks roughly toward the camera. Restore or upscale a faded or blurry print first, crop to head and shoulders, then animate it. Only do this with photos your family is happy to see animated.

Link to this answer

Can I make two people in the same photo talk?

Yes. InfiniteTalk Duo, InfiniteTalk Fast Duo and LongCat 1.5 Duo take a photo of two people and one audio track for each, played together or in turn, so each person speaks their own lines.

Link to this answer

How long can the talking video be?

Up to 10 minutes of speech on Kenerate Avatar, InfiniteTalk and Kling Avatar 1 Pro; 5 minutes on Kling Avatar 2, Sync Lipsync 3 and VEED Fabric; 60 seconds on OmniHuman 1.5; 5 to 20 seconds on LTX Lipsync.

Link to this answer

Why does my talking photo look strange?

Usually the photo: a turned head, an open mouth, sunglasses or a very small face. Music under the voice also confuses the lip sync. Fix those first; if it still looks off, try Sync Lipsync 3 for the mouth or OmniHuman 1.5 for more natural movement.

Link to this answer

Can I make a celebrity's photo talk?

No. Only animate photos of yourself, characters you made, or people who agreed to it.

Link to this answer

What the engine actually does

A talking-photo engine reads the audio, works out the mouth shape for each sound and redraws the face frame by frame to match. Good engines also add blinks, small head turns and shoulder movement, so the picture doesn't look like a cut-out with a moving mouth. Some, such as OmniHuman 1.5, animate the whole body.

Explore

Every Kenerate video tool

Generators, tools, models and guides — all in one studio.

See more of Kenerate AI in your Google resultsAdd kenerateai.com as a preferred source — Google shows more from your preferred sources in Top Stories and AI answers.

By Last updated and reviewed

Sources: Kenerate Avatar Talk · Kenerate Voice Studio · OmniHuman-1 project page

Make your photo talk

One studio for images and video.

Tried it? Tell us how it went