Ijharul Islam
To make a photo talk with AI, upload a clear front-facing portrait to a talking-photo tool, give it speech (a typed script it voices, or your own audio file), and let the model animate the mouth, face and head to match. Most tools return a short video in a few minutes.
The quality depends mostly on the photo you start with and the model you pick. This guide explains both, shows the steps and costs in QuickVeed, lists free open-source options, and covers the consent rules, which matter a lot with real people's faces.
The short answer
Pick a sharp, well-lit, front-facing photo with the mouth closed and the face unobstructed. Upload it to a talking-photo tool, add a typed script or a clean audio file, choose a model, and generate. Use only photos of yourself or people who agreed, and label realistic results as AI where platforms ask.
Key takeaways
- The photo decides most of the quality: front-facing, sharp, even light, nothing covering the mouth.
- Script-driven tools voice your text; audio-driven tools move the face to a recording you supply.
- Cheaper models give a quick, lighter animation; higher-end models look more natural but cost more.
- Never animate a real person without their permission, and don't make anyone appear to say something they didn't.
Disclosure: we make QuickVeed and show how to do this in it, along with other ways. Details about other products come from their public pages, checked in October 2026 and linked at the end.
How talking-photo AI works
A talking-photo model takes one still image and a speech track and predicts how that face would move while saying those words: lip shapes, jaw, blinks and small head movements. Some models only animate the face. Others generate the head, shoulders and hands too, which looks more like a real video.
There are two ways to supply the speech:
- Script-driven: you type the words and the tool generates both the voice and the animation.
- Audio-driven: you upload a recording (your voice, a voiceover, or text to speech made elsewhere) and the face is animated to match it.
Audio-driven gives you full control of the voice, including a cloned voice of your own. Script-driven is faster when you don't have audio yet.
Common uses for a talking photo:
| Use | Typical length | Tip |
|---|---|---|
| Website welcome from the founder | 20-30 s | Use your own recorded voice for trust |
| Social post with a character or mascot | 10-20 s | Illustrated faces need a clear, front-facing mouth |
| Product explainer with a presenter | 20-30 s | Keep the script to one benefit |
| Teaching history or language with a portrait | 15-30 s | Label it as an AI reconstruction |
| Personal greetings and invitations | 10-20 s | Only with the person's permission |
Choosing a photo that animates well
| Works well | Causes problems |
|---|---|
| Face looking at the camera | Strong side profile |
| Mouth closed or relaxed | Wide smile or open mouth |
| Even, soft light | Hard shadows across the face |
| Sharp, high-resolution original | Small, blurry or heavily compressed images |
| Head and shoulders in frame | Face cropped at the chin or forehead |
| Nothing in front of the face | Hands, microphones, hair or glasses glare over the mouth |
Illustrations and paintings can work, but stylised faces sometimes warp. Try a few seconds first before generating a full clip.
Step by step
- Write the script or record the audio. Keep it short: many tools limit a clip to 30 seconds or so. About 75 words fills 30 seconds.
- Prepare the photo. Crop to head and shoulders, fix exposure, and use the highest resolution you have.
- Pick a model. Start with a cheap one to test the photo, then move up if you need more natural motion.
- Generate a short test. Check the lips against the sound, the teeth, and the eyes.
- Generate the full clip and add captions if it's going on social media.
Preparing the audio
If you use an audio-driven model, the recording matters as much as the photo. Record in a quiet room, close to the microphone, with no background music. Trim silence at the start and end so the face doesn't sit frozen, and keep a natural pace: very fast speech gives blurry mouth shapes. If the audio is from a video, our free extract audio from video tool pulls it out in your browser.
Common problems and fixes
| Problem | Fix |
|---|---|
| Mouth looks smeared | Use a sharper photo, or a higher-end model |
| Teeth look strange | Start from a photo with the mouth closed |
| Head barely moves | Try a model that animates the head as well as the mouth |
| Face drifts or changes identity | Shorten the clip and use a front-facing photo |
| Lips are out of sync | Clean the audio and remove long silences |
Making a photo talk in QuickVeed
QuickVeed's AI avatars offer two ways to animate a photo, up to 30 seconds per video, with one of 22 stock avatars or your own photo:
| Mode and model | What it does | Credits per 10 s |
|---|---|---|
| Photo to Talking Video: LivePortrait | Quick, light animation of a portrait | 10 |
| Photo to Talking Video: SadTalker | Animates a portrait's face and head to speech | 21 |
| Full Avatar: P-Video Avatar | Speaks a script you type | 63 |
| Full Avatar: Omni-Human | Animates a face to your own audio, with more natural motion | 350 |
- Open AI avatars in the dashboard.
- Choose Photo to Talking Video for a cheap, quick result, or Full Avatar for a more natural presenter.
- Upload your photo (or pick a stock avatar).
- Type the script, or upload the audio if the model is driven by audio.
- Generate. The video is saved in your asset library.
A 30-second clip costs 30 credits with LivePortrait, 63 with SadTalker, 189 with P-Video Avatar and 1,050 with Omni-Human. The 75 trial credits are enough to try the cheaper models on your own photo. If you want your own voice, record it, or clone it with the AI voice generator first and use the audio with Omni-Human.
Free and other options
- SadTalker is open source under the Apache 2.0 licence and turns a single portrait plus an audio file into a talking-head video. It runs on your own computer, ideally with a recent graphics card, and needs some comfort with Python.
- LivePortrait, from Kuaishou's research team, is open source and animates a portrait by following the motion of a driving video, so you can act out the expression yourself.
- HeyGen offers a talking-photo feature alongside its avatars, in many more languages than QuickVeed.
Phone apps also offer talking-photo effects. Check their terms before using them for business: some keep rights to what you upload.
Consent and labelling
A talking photo can make anyone appear to say anything, so set rules for yourself before you start:
- Animate only your own face, a stock avatar, or someone who has agreed in writing.
- Never put words in a real person's mouth that they didn't say, including as a joke about a public figure.
- Be careful with photos of people who have died. Ask the family first.
- Label realistic results. YouTube asks creators to disclose realistic altered or synthetic content, and TikTok requires a label on realistic AI-generated video.
Frequently asked questions
Can I make a photo talk for free?
Yes, with open-source tools like SadTalker if you can run them on your own computer. Many online tools also include a few free credits; QuickVeed's 14-day trial gives 75 credits, enough for several short tests.
Which AI makes the most realistic talking photo?
Higher-end models such as Omni-Human usually look more natural than quick portrait animators, but they cost much more. Test on your own photo before choosing.
How long can a talking photo video be?
It depends on the tool. QuickVeed makes up to 30 seconds per video; join several clips for longer content.
Can I use my own voice for a talking photo?
Yes. Upload a recording to an audio-driven model, or clone your voice and generate the audio from text.
Is it legal to animate someone's photo?
Laws vary by country, but using someone's likeness without consent can break privacy, publicity or fraud laws and platform rules. Get permission and never use it to deceive.
Try it with your own video
Try a 10-second test on your own photo. Describe the video or paste your script, pick a voice, and judge the first draft for yourself. QuickVeed plans start at $18/month, and new accounts get a 14-day trial with 75 credits, no card needed. Sign up or see pricing.
Sources
Competitor details in this post were checked against these public pages in October 2026. Products and prices change, so confirm with each vendor before you decide.
