Talking avatar videos on Pixel Dojo
P Video Avatar: AI Talking Avatar Generator
P Video Avatar turns one portrait and a script into a talking video that says every word you wrote, in the voice and mood you asked for.
P Video Avatar animates a single portrait into a talking head video. Type a script and pick one of 30 voices in 10 languages, or upload your own audio, and it lip syncs the face with natural nods, smiles and blinks. It works on photos, 3D characters and painted illustrations alike, at 720p or 1080p. We made every example on this page on Pixel Dojo, starting from portraits we generated, and each one comes with its exact script and voice settings.
Already a member? Sign in
By Brian Lovett · Updated
How P Video Avatar turns a still portrait into a talking video
You give P Video Avatar two things: a portrait and something to say. The portrait can be a photo, a 3D render or a painting. What to say can be a typed script, voiced by one of 30 built in voices, or an audio file you upload, which the face lip syncs to.
Two short directions shape the performance. The voice direction sets the read, like "Low, gravelly film noir narration, slow and dry". The video direction sets the body language, like "Knowing smile, a slight lean toward the camera at the end". Our detective barely moves and blinks slowly. Our innkeeper leans in for her last line.
We checked every soundtrack on this page with a speech to text pass, and each one came back word for word: the English, the French and the Spanish scripts alike.
Start from a portrait facing the camera, centered from the chest up, with the mouth closed or in a soft smile. That is how we framed all six portraits we made for these examples.
P Video Avatar examples: presenters, hosts and characters
Five talking videos at 1080p, each made from one still portrait and a short script. Unmute a tile to hear it, then tap Use this prompt to start from the same script.
Product presenter, English (US), 12 seconds. An upbeat, confident read with small nods: a ready made spokesperson clip for a product page.
Meet the desk lamp that follows the sun. It shifts from cool morning light to a warm evening glow on its own, so your workspace always feels right.
Podcast host, English (US), 12 seconds. Relaxed late night delivery with head tilts and glances toward the mic.
Welcome back to Night Shift Notes. Today we are talking about why the best ideas show up at two in the morning, and how to catch them before they disappear.
3D character, English (UK), 12 seconds. A feature animation style portrait that talks like a favorite teacher sharing a secret.
Ferns love a little humidity. Mist the leaves every morning, keep them out of direct sun, and they will reward you with new fronds every single week.
Painted illustration, English (US), 15 seconds. Low, gravelly noir narration over a comic book panel that comes to life.
The rain had not stopped in three days, and neither had the phone. Somebody in this city knew where the necklace went, and I intended to find them.
Fantasy character, English (UK), 15 seconds. A wry, conspiratorial line that drops to a whisper at the end, ready for a game trailer or an RPG intro.
Welcome, traveler. Rooms are three silver a night, the stew is hot, and if you are looking for the old map, you did not hear it from me.
One avatar, many languages: French and Spanish tests
P Video Avatar speaks 10 languages. We gave our presenter the same product script in French, and wrote a quick cooking tip in Spanish for a vertical 9:16 home cook. Both came back with every word of the script and lip sync that matches the language.
French, 9 seconds at 1080p. The same face and voice as the English presenter, now selling the lamp in French.
Voici la lampe qui suit le soleil. Elle passe d'une lumière fraîche le matin à une lueur chaude le soir, toute seule.
Spanish, 9:16 vertical, 11 seconds at 1080p. A lively phone style tip for Reels and TikTok.
¡Hola! Mi truco para una salsa perfecta es un chorrito de limón al final. Despierta todos los sabores. Pruébalo esta noche y me cuentas.
- Write the script in the target language and set the matching voice language. Our French and Spanish clips speak the script exactly.
- Reuse one portrait for every market: the same presenter sells in English and French.
- Use a vertical portrait for a vertical video. The 9:16 cook came from a 9:16 portrait.
- Keep scripts to a few short sentences. Our clips run 9 to 15 seconds, the right length for a product card or a social post.
P Video Avatar settings and inputs
Everything you set for a talking video, with the values we used.
- Portrait imagejpg, png or webp
- One face, looking at the camera, chest up and centered. Photos, 3D renders and illustrations all work.
- Voice scriptAny text
- What the avatar says. Write it the way people talk, in short sentences.
- Voice30 named voices
- Pick a voice that fits the character: we used Autonoe for the presenter, Umbriel for the podcast host and Algenib for the detective.
- Voice language10 languages
- English (US) and English (UK), French and Spanish are all on this page.
- Voice directionFree text
- How to say it: "Relaxed, warm late night podcast delivery with a hint of humor."
- Video directionFree text
- How to move: "Relaxed expression, slight head tilts, glances toward the microphone, steady camera."
- Audio uploadOptional
- Upload your own voice or a recorded line instead of a script, and the avatar lip syncs to it.
- Resolution720p, 1080p
- Our 1080p clips measure 1920 x 1088 landscape and 1088 x 1920 vertical.
P Video Avatar pricing: 1 credit a second at 720p
P Video Avatar is included in every Pixel Dojo plan, from $10/month. It costs 1 credit per second of video at 720p and 2 credits per second at 1080p.
For a typed script, the length is estimated from the word count before you generate, so you see the price up front. Our 12 second 1080p presenter cost 24 credits.
We made every clip on this page through the Pixel Dojo API, and the same scripts cost the same in the web tool.
P Video Avatar questions
What is P Video Avatar?
P Video Avatar is an AI talking avatar model. It animates one portrait into a talking head video, driven by a typed script with a built in voice or by an audio file you upload, with lip sync and natural head movement.
Can I make a talking video from a photo?
Yes. Upload a portrait that faces the camera, type what it should say and pick a voice. Every example on this page started as a single still image.
Does it work with cartoons, 3D characters and illustrations?
Yes. Our 3D botanist, graphic novel detective and painted innkeeper all talk with matching lip sync and expressions.
Which languages and voices are available?
There are 30 named voices across 10 languages, including English (US), English (UK), French and Spanish. You can also upload your own audio instead of a script.
How much does P Video Avatar cost?
It is included in every plan, from $10/month, and costs 1 credit per second at 720p and 2 credits per second at 1080p. A 12 second 1080p clip like our presenter is 24 credits.
Can I use P Video Avatar through an API?
Yes. POST an image URL with a voice_script and voice, or an audio URL, to /api/v1/models/p-video-avatar/run. Every clip on this page was made through the Pixel Dojo API.
Make your first P Video Avatar clip
Write a prompt above or start from any example on this page. One subscription covers every model on Pixel Dojo.
Start with P Video Avatar