1. Pick a photo
Sign in and choose a photo of one person, then confirm it is you or that they agreed.
OmniHuman v2
01 · Audio-driven
The sound moves the face.
OmniHuman v2 takes a single photo and an audio track and draws the person speaking it. The lips follow the words; the head, the brows and the hands follow how they are said.
02 · Speech or song
It talks, and it sings.
A voiceover, an interview answer or a verse of a song all drive it the same way. The clip is exactly as long as the audio.
- On Phantra, 4 seconds and up, to 30 seconds on Standard, 60 seconds on Pro, 120 seconds on Max.
03 · Direction
A line of direction sets the shot.
An optional prompt sets the scene, the movement and the camera: a slow push in, a turn to the window, a laugh at the end.
What OmniHuman v2 Takes
- 01
One photo
A still of one person, 300 to 6000 pixels a side, from a tall portrait to a wide frame.
- 02
One audio track
A recording you upload (.mp3 or .wav), or a script the studio reads aloud for you first.
- 03
A typed script
On Phantra, a script is voiced by ElevenLabs v3 in one of 21 voices, and you hear it before the clip.
- 04
Lips, face and hands
Mouth shapes, expressions, head turns and gestures, all timed to the sound.
- 05
An optional prompt
Scene, movement and camera in a line. Written in English, Chinese, Japanese, Korean, Spanish or Indonesian.
- 06
As long as the audio
The model takes up to 300 seconds; Phantra makes 30 seconds on Standard, 60 seconds on Pro, 120 seconds on Max.
- 07
Priced by the second
Billed on the length of the voice, rounded up to the whole second, and shown on the card before it runs.
- 08
Your own photos
An upload, a still you made in the studio or one of your saved elements.
- 09
Then the editor
The clip is a render like any other: upscale it, reframe it or cut it with other clips.
OmniHuman v2 Pricing
What an OmniHuman v2 clip costs on Phantra, in credits. One rate per second of voice, so the price depends only on its length.
| Clip | Credits |
|---|---|
| 10-second clip | 252 |
| 30-second clip | 756 |
| 60-second clip | 1,512 |
| 120-second clip | 3,024 |
- The longest voice per clip is 30 seconds on Standard, 60 seconds on Pro, 120 seconds on Max.
- A typed script's voice preview starts at 10 credits; an uploaded recording costs nothing extra.
- ByteDance also sells OmniHuman through its own apps and partners; we quote none of their prices here. Link checked 2026-10-20.
How to use OmniHuman v2
2. Give it a voice
Upload a recording, or type a script, choose a voice and listen to the preview.
3. Approve the clip
The card shows the credits for the voice's length. Approve it and the clip lands in your Library.
FAQs about OmniHuman v2
OmniHuman v2
One photo. One voice. A clip of them saying it.
Type the words or upload them, hear the voice, then approve the clip at the price on the card.