AI avatar model
Bring a single portrait to life with OmniHuman 1.5 — upload a photo and an audio clip (or type text and pick a voice) to generate a realistic talking avatar with synced lips, expressions, and gestures.
Compare OmniHuman 1.5 with other video models — same workspace, one click to switch.
A glimpse of what people are creating with AI video models on CreateVision AI.
ByteDance's OmniHuman 1.5 drives a single portrait from audio: synced lips, natural expressions, and whole-body gestures that respond to the rhythm and meaning of the voice — not just a moving mouth. Upload a photo and a clip, or type text and pick a voice.
Give OmniHuman 1.5 a single portrait and an audio clip and it animates the face into a believable speaking performance with accurate lip-sync, while preserving the person's identity, facial structure, and lighting. No rigging, no green screen, no studio.
Unlike basic talking-head tools, OmniHuman 1.5 reads the rhythm, prosody, and meaning of the audio to drive head movement, posture shifts, and hand gestures — so the avatar emphasizes, pauses, and reacts like a real presenter rather than a static portrait with a moving mouth.
No audio on hand? Switch to text mode: type what the avatar should say, choose a voice from the built-in text-to-speech library, and OmniHuman 1.5 generates the spoken performance for you. Ideal for scripted presenters, explainers, and social clips.
The model adapts to different audio styles — calm narration, expressive dialogue, even singing — and can follow optional text guidance for camera and gesture direction, producing polished results suitable for marketing, training, and social content.
One account, the full image and video generation workflow.
Start free with daily credits. Upgrade for premium models, faster generation, and 4K output.
Perfect for getting started
Billed annually · $240/year
Billed annually · $300/year
OmniHuman 1.5 is ByteDance's audio-driven avatar model: it turns a single portrait plus a voice track into a realistic talking video with synced lips, expressions, and gestures.
A clear single-person portrait photo, plus either an audio clip (mp3/wav) or — in text mode — the text you want spoken and a voice picked from the built-in library, which is synthesized into the driving audio.
Up to 35 seconds of audio per generation. Longer scripts are best split into multiple clips. Billing is by the second of audio, rounded up.
Yes. OmniHuman 1.5 generates head motion, posture, and hand gestures that respond to the audio's rhythm and meaning, so the avatar performs rather than just lip-syncs.
A single, clearly visible, front-facing person with good lighting and an unobstructed face. Group photos, tiny faces, or heavily occluded faces reduce quality or may fail.
Per second of generated audio. The exact credit cost is shown before you generate, and failed runs are refunded automatically.
