AI Lip Sync Models

5 models available on BudgetPixel

BudgetPixel has five AI lip sync models in the Video Workshop's Lip Sync tool. Three make a talking video from one photo and a voice: OmniHuman 2.0 by ByteDance, the newest, with up to 2 minutes of audio and a prompt for expression and camera, at 84 credits a second; OmniHuman 1.5, up to 60 seconds at 720p or 1080p; and P-Video Avatar by Pruna AI, the low-cost pick for clips up to 30 seconds. Two re-sync the mouth in a video you already have to new audio: Sync Lipsync 2.0 and 2.0 Pro by Sync Labs.

Every lip sync model on BudgetPixel. They do one of two jobs. Photo-to-video models (OmniHuman 2.0, OmniHuman 1.5 and P-Video Avatar) animate a single picture to a voice: the lips follow the words, and the face, head and hands move with the speech. Video re-sync models (Sync Lipsync 2.0 and 2.0 Pro) keep a video you already have and change only the mouth, to match new audio for dubbing or a rewritten line.

The voice can be any recording, or a line typed in the workshop and spoken by a built-in voice or your own cloned voice, which is free. Every model is priced per second of audio (or of video, for the re-sync models), and the Lip Sync tool shows the price before you start.

Open Lip Sync

All lip sync models

  • OmniHuman 2.084 credits/sec

    A photo and a voice become a talking video: lips, face, gestures and camera, directed with an optional prompt. Up to 2 minutes at 720p.

  • P-Video Avatar40-60 credits/sec

    A fast, low-cost talking avatar from one photo and a speech clip, at 720p or 1080p.

  • OmniHuman 1.5135 credits/sec

    The earlier OmniHuman: a talking video from a photo and audio, with an optional prompt, at 720p or 1080p.

  • Sync Lipsync 2.055 credits/sec

    Re-syncs the mouth in a video you already have to new audio, for dubbing and voice changes.

  • Sync Lipsync 2.0 Pro90 credits/sec

    The higher-detail Sync re-sync, for close-ups and faces that fill the frame.

AI lip sync models compared

AI lip sync models compared
FeatureYou give itLengthOutputPrice
OmniHuman 2.0A photo + audioUp to 120 s of audio (4 s minimum)720p, in the photo's shape84 credits/sec
P-Video AvatarA photo + audioUp to 30 s of audio720p or 1080p40-60 credits/sec
OmniHuman 1.5A photo + audioUp to 60 s of audio720p or 1080p135 credits/sec
Sync Lipsync 2.0A video + audioUp to 60 s of videoYour video, new mouth movement55 credits/sec
Sync Lipsync 2.0 ProA video + audioUp to 60 s of videoYour video, new mouth movement90 credits/sec

Photos that talk, made with OmniHuman 2.0

  • An old fisherman on a harbor at dawn, talking to the camera in an OmniHuman 2.0 video
    A FLUX 3 Video frame, speaking English
  • A woman with curly dark hair at a sunny café table, talking in Spanish in an OmniHuman 2.0 video
    A portrait, speaking Spanish
  • A presenter in a navy blazer explaining with open hands in an OmniHuman 2.0 video
    A presenter, directed with a prompt
  • A golden retriever in a garden, talking excitedly in an OmniHuman 2.0 video
    Even a dog

AI Lip Sync Models — common questions

What is AI lip sync?
A model that makes a mouth move in time with speech. Some animate a still photo into a talking video; others take a video that already exists and redraw the mouth to match new audio, which is how videos are dubbed into other languages.
Which lip sync model should I use?
To make a photo talk, OmniHuman 2.0: it takes the longest audio (2 minutes), follows a prompt for expression, gestures and camera, and costs less per second than OmniHuman 1.5. P-Video Avatar is the budget pick for clips up to 30 seconds, and OmniHuman 1.5 or P-Video Avatar give 1080p. To change what someone says in an existing video, Sync Lipsync 2.0, or 2.0 Pro for close-ups.
Can I make any photo talk?
A photo with one clearly visible face, looking roughly at the camera, works best: a portrait, a half-body shot, an illustrated character or even a pet. For OmniHuman 2.0 the picture's aspect ratio must be between 2:5 and 5:2; any size works, as small pictures are scaled up and very large ones down.
What audio can I use?
Speech in any language, as an MP3 or WAV file or recorded in the browser. You can also type a line and have it spoken by one of BudgetPixel's voices or your own cloned voice, for free. OmniHuman 2.0 takes up to 120 seconds of audio, OmniHuman 1.5 60 and P-Video Avatar 30; the Sync models re-sync videos up to 60 seconds long.
How much does AI lip sync cost?
Per second: OmniHuman 2.0 84 credits (4-second minimum), OmniHuman 1.5 135, P-Video Avatar 40 at 720p or 60 at 1080p, Sync Lipsync 2.0 55 and 2.0 Pro 90. The price is set aside when the job starts and charged only if it succeeds.
How long does it take?
A few minutes. In our tests a 6-second OmniHuman 2.0 clip took about 3 minutes, P-Video Avatar about 1 minute and Sync Lipsync about 2.
Can I use lip sync through the API?
Yes, for the photo models: OmniHuman 2.0, OmniHuman 1.5 and P-Video Avatar are in the BudgetPixel API under POST /v1/lip-sync/{model}, billed the same way as in the workshop. The Sync models are in the workshop only.

Browse other model types

  • AI Image Generation Models
  • AI Video Generation Models
  • AI Music Generation Models
  • AI Voice & Text-to-Speech Models
  • AI 3D Model Generators
  • AI Video Upscalers
  • All AI models