Characters · Lesson 5 of 6
How to make a talking avatar from a photo
One portrait and one line of text become a talking video. OmniHuman writes the voice for you, then brings the face to life with lip sync and gestures.
You need: One clear, front-facing portrait and the line they should say. · Start to finish, step 8 of 10
Step by step
Open OmniHuman and add a portrait
Click Add portrait and pick a clear, well-lit, front-facing photo. A saved character works best: here, a close-up of Juniper made in Character Studio.
Make the voice with Text to speech
No recording needed. Click Text to speech, type the line, pick a voice and an emotion, then press Generate audio. Have a recording already? Upload it instead.
Describe how they move
"Warm and smiling, talking to the camera with small hand gestures." Optional, and worth it: it sets the performance.
Generate and save
The price follows the length of the audio. About a minute later they say the line with natural lip sync. Press Save to keep it.
Why it works
- The audio drives everything. A natural read with pauses makes a natural performance.
- Use a saved character portrait to give a recurring presenter the same face every episode.
Made this way
Your practice task
Make a 10-second welcome message for your own channel with a saved character.
Read the transcript(1:49)
Turn one portrait and one line of text into a talking video with OmniHuman's built-in text-to-speech. Open OmniHuman in PixelDojo. It combines a portrait with a voice clip to create a talking, moving character. We'll make that voice from text without recording anything. Click Add Portrait and choose a clear, front-facing photo. For this example, we're using a close-up of Juniper, a character made in Character Studio, as our starting portrait. Click Text-to-Speech, then enter Juniper's line. Hi, I'm Juniper. This whole video started as one photo and one line of text. That's the dialogue she'll speak. Choose a voice from the Voice menu. Here, we're using Friendly Person with the Emotion set to Neutral. Press Generate Audio to turn the written line into speech. Once the audio is ready, it lands directly in the composer beneath the portrait. Juniper now has her spoken line, and the audio player shows its length. Add the movement description. Warm and smiling, talking to the camera with small hand gestures. The price beside Generate follows the length of the audio you've created. With the portrait, voice, and movement description in place, press Generate. OmniHuman uses those inputs to make Juniper say the line, with natural lip-sync and gestures. About a minute later, the talking video is ready. Juniper speaks the line with the requested expression and movement. Click Save All to keep your generated video. Hi, I'm Juniper. This whole video started as one photo and one line of text. You've turned a portrait and text into a talking video. Next, try another line with a different voice.