video model · by ByteDance
OmniHuman 1.5 AI Talking-Head Generator
Animate a portrait photo with audio to produce a realistic talking-head video. OmniHuman 1.5 is ByteDance's lip-sync video model - upload a still portrait and an audio clip, get back a fully animated talking head with synchronized mouth, expression, and head motion.
Generated with OmniHuman 1.5
A handful of OmniHuman 1.5 clips generated inside Fliki. No edits, no post.
Prompt
Prompt
Prompt
Trusted by 50,000+ companies worldwide
What makes OmniHuman 1.5 the right talking-head model
Photo + audio talking head
OmniHuman 1.5 takes a single portrait image and an audio clip and produces a realistic talking-head video. No video reference needed — just a photo and the audio you want spoken.
Tight lip-sync accuracy
Mouth shape and timing align with the audio at phoneme level — the kind of sync quality you'd otherwise need a recorded performance for.
Natural micro-expressions
Beyond mouth motion, OmniHuman 1.5 generates eye movement, blinks, and subtle head motion — the small cues that make talking-head video feel alive rather than uncanny.
Identity retention
OmniHuman 1.5 holds the source portrait’s identity across the full clip — likeness, styling, and lighting carry through without drift.
10-second clips at 720p
OmniHuman 1.5 outputs 10-second 720p talking-head clips. Stitch multiple generations together inside Fliki to build longer presenter sequences.
No video source required
Unlike face-swap or motion-transfer workflows, OmniHuman 1.5 doesn't need an existing video clip. A single photo and an audio file are enough — that's a much lower asset bar for production.
Built for presenter content
Founder-led explainers, multilingual versions of a single recording, personalized outreach, and presenter cuts for ads and social — all from a single source portrait.
Inside Fliki, no setup
Run OmniHuman 1.5 through Fliki's standard interface — no separate ByteDance account, no Runware setup. Upload, attach audio, generate.
Works across visual styles
OmniHuman 1.5 handles realistic photographs, anime characters, illustrated portraits, and stylized artwork — plus non-human subjects like animals and anthropomorphic figures. The same model carries a brand mascot or a real spokesperson.
How it works
How to make a talking-head video with OmniHuman 1.5
OmniHuman 1.5 runs inside Fliki's lip-sync workflow. Five steps.
Upload a portrait image
Use a clear, front-facing portrait. Better source clarity gives the model more to work with — sharp lighting, unobstructed face, neutral expression all help.
Select OmniHuman 1.5
Pick OmniHuman 1.5 from Fliki's lip-sync model selector. It's designed specifically for image-plus-audio talking-head workflows.
Upload your audio
Upload the dialogue or voiceover audio you want the portrait to speak. OmniHuman 1.5 syncs mouth movement, micro-expressions, and head motion to the audio.
Generate at 720p
Hit Generate. OmniHuman 1.5 produces a 10-second 720p talking-head clip with the source identity and the new spoken content.
Drop into your timeline
Use the output as a presenter clip, an explainer talking head, a personalized message, or a social hook. Inside Fliki you can stitch multiple OmniHuman clips together for longer presenter sequences.
OmniHuman 1.5 FAQ
Frequently asked questions
Everything you need to know about generating talking-head video with OmniHuman 1.5 inside Fliki.
What is OmniHuman 1.5?
OmniHuman 1.5 is ByteDance's lip-sync video model. It takes a single portrait image and an audio clip, and produces a realistic talking-head video with synchronized mouth movement, expressions, and natural head motion.
What inputs does OmniHuman 1.5 need?
A single portrait image and an audio file. No source video required, no text prompt — the audio drives the spoken content and timing.
How long are OmniHuman 1.5 outputs?
OmniHuman 1.5 generates 10-second talking-head clips. Stitch multiple generations together inside Fliki to build longer presenter sequences.
What resolution does OmniHuman 1.5 output?
OmniHuman 1.5 generates at 720p, optimized for talking-head quality and lip-sync sharpness.
How is OmniHuman 1.5 different from a regular video model?
It’s lip-sync only — designed specifically for the photo-plus-audio talking-head workflow. Regular text-to-video and image-to-video models like Veo 3.1 Fast or Kling 3.0 Pro generate full scenes; OmniHuman 1.5 specializes in animated portraits driven by audio.
Can I use any portrait photo?
Most clear, front-facing portraits work well. Sharp lighting, unobstructed face, and neutral expression give the model the strongest source to work with.
Still curious?
Try Fliki free in your browser, no credit card required.
Start freeCan OmniHuman 1.5 work in any language?
Yes. The model syncs to whatever audio you provide — English, Spanish, Mandarin, anything. Pair it with Fliki’s multilingual voices to localize a presenter recording in minutes.
Can I use OmniHuman 1.5 for commercial content?
Yes, subject to Fliki's and ByteDance's terms, plus the rights you hold to the source portrait. Always make sure you have permission to use someone's likeness in commercial contexts.
AI video models
Discover more
Tools
Discover features
Turn a photo into a talking head.
Upload a portrait, attach audio, get a realistic 720p talking-head clip. Free to start, no credit card required.
Try OmniHuman 1.5 freeFree forever plan · No credit card required · Cancel anytime