Talking Photo AI: Make Any Portrait Speak
A talking photo on Wandura turns one portrait and one voice recording into a video where the face speaks your audio with synced lips and natural head motion. Upload a portrait and an MP3, WAV or M4A voice file, pick 720p or 480p, and Wandura generates the clip; the credit price is shown on Create before you start.
Make Any Photo Talk in a Few Clicks
Wandura Talking Avatar takes exactly two inputs - a portrait image and a voice audio file - and generates a video where the face speaks every word in sync. Two real engines power it: VEED Fabric 1.0, the default, and OmniHuman 1.5 as the alternative. Each run takes a short voice clip: up to 10 seconds on VEED Fabric 1.0, up to 9 seconds on OmniHuman 1.5. There is no script field and no prompt to write; the avatar says whatever your audio says. Each model shows its price in the widget before you create, and the finished video (720p, or 480p on VEED Fabric) is saved straight to your Creations.
Try AI Talking AvatarPaid plans only · from $15/month
A Presenter From a Photo and a Script
An AI talking avatar turns a portrait plus a script or audio track into a presenter who delivers your message on camera — no studio, no talent, no reshoots. It is built for explainers, spokesperson ads, and localized content at volume.
Use a front-facing portrait and clear audio (or generate the voice with text-to-speech) for the most natural delivery. Swap the script to produce the same presenter speaking different offers or languages.
Talking Photo at a Glance
| Item | Details |
|---|---|
| Inputs | One portrait and one voice file (MP3, WAV or M4A); there is no script or prompt field |
| VEED Fabric 1.0 (default) | Audio up to 10 seconds; 720p or 480p |
| OmniHuman 1.5 | Audio up to 9 seconds; 720p |
| Price | Shown on Create before you start (1 credit = $0.01) |
2 AI Models, One Talking Avatar Tool
A Presenter from Nothing but a Voiceover
Have a script recorded but no on-camera talent? Upload a portrait and your voice track and Wandura turns them into a talking presenter for explainers, course intros and product walkthroughs - reshoot-free updates whenever the script changes.
Try AI Talking AvatarFaceless Channels, with a Face
Run a content channel without showing your own face: pair one consistent portrait with fresh audio for every episode. The same photo becomes a recurring host your audience recognizes across videos - the clip here is a real result: the same portrait as above, speaking a fresh audio track.
Try AI Talking AvatarBring Old Photos and Mascots to Life
Animate a family photo, a historical portrait or a brand mascot with a recorded message - greetings, tributes and social moments made from a single still image.
Try AI Talking AvatarSpeak Every Language You Can Record
The avatar speaks whatever is in the audio - so chain in Text to Speech or Voice Clone output to localize one portrait into any language, without the speaker learning a word of it.
Try AI Talking AvatarBrand Characters That Announce Things
Give product updates, promos and campaign messages a consistent spokesperson: one approved portrait plus a new voice track per announcement, iterated in minutes instead of studio days.
Try AI Talking AvatarHow It Works
- Step 01
Step 1: Upload a portrait
Add a clear, front-facing photo with the face fully visible - the sharper the portrait, the more natural the animation.
- Step 02
Step 2: Upload a voice track
Add the audio the avatar should speak - a recording, a Voice Clone result, or a Text to Speech export. Keep it within 10 seconds on VEED Fabric 1.0 or 9 seconds on OmniHuman 1.5.
- Step 03
Step 3: Create
Pick a model (and on VEED Fabric 1.0, 720p or 480p) and hit Create. The talking video is saved to your Creations, ready to download or chain onward.
How Much Does AI Talking Avatar Cost?
Every model below runs on one Wandura credit balance — pay only for what you generate, with no separate subscription per model.
| Model | Pricing |
|---|---|
| VEED Fabric 1.0 | Default · audio ≤10s |
 OmniHuman 1.5 | Audio ≤9s |
FAQs
It takes a portrait photo and a voice track and generates a video where the face speaks in sync with the audio - mouth movement, head motion and expression all follow the voice.
Exactly two things: a portrait image and a voice audio file. There is no script or prompt field - the avatar speaks whatever is in the audio, in whatever language you recorded.
A clear, well-lit, front-facing photo with one face fully visible and unobstructed - no sunglasses, heavy shadows or extreme angles. Both models are built around a detectable face, so a photo without a clear face will not animate properly.
Up to 10 seconds per run on VEED Fabric 1.0 and up to 9 seconds on OmniHuman 1.5; the tool tells you before you create if a clip is longer. For a longer script, split it into short clips and generate each one. Clean, clearly spoken speech gives the most accurate lip movement on both.
VEED Fabric 1.0 is the default; OmniHuman 1.5 is the alternative. Both take the same two inputs; VEED Fabric renders at 720p or 480p and OmniHuman 1.5 at 720p - switch between them per run and compare.
The price depends on the model, resolution and audio you add, and the Create button shows the exact credits before you start (1 credit = $0.01). Credits are only deducted when you generate.
Talking Avatar renders at up to 720p: VEED Fabric 1.0 offers 720p or 480p, and OmniHuman 1.5 renders at 720p. Need it sharper? Chain the result into Video Upscaler.
Generation runs asynchronously in the background - you can leave the page, and the finished video appears in your Creations once it is ready.
Talking Avatar starts from a still photo plus audio and animates the whole face from scratch. Lip Sync starts from an existing face video plus audio and re-times the mouth in footage that already moves. Only have a portrait? Use Talking Avatar. Have footage? Use Lip Sync.
Record it yourself, or generate it with Text to Speech or Voice Clone and upload the audio file. The avatar speaks whatever is in the audio, in any language you can record.
Each run takes up to 10 seconds of audio on VEED Fabric 1.0 and up to 9 seconds on OmniHuman 1.5. For a longer script, split it into short clips and generate each one.
