Loading the page…
OmniHuman 2.0 by ByteDance: brings a photo to life with your voice. Face, gestures and lips follow audio up to 5 minutes.
OmniHuman 2.0 is available in IIshki Studio — video generation from a prompt, paid for with tokens. IIshki club members get reduced prices on every model. Pick a model above, set the parameters and start the generation.
OmniHuman 2.0 is ByteDance's talking-avatar model: from one photo and an audio track it makes a clip where a person, character or pet talks or sings along to your sound. Face, gestures and lips follow the voice; the clip is as long as the audio, up to 5 minutes.
Speaks calmly and warmly, nods slightly, gestures with both hands
Sings with eyes closed, swaying to the music
MP3 or WAV from 4 seconds to 5 minutes, up to 15 MB. A voice recording, a song or a voice-over from the IIshki audio studio all work.
The prompt is optional. It refines behaviour and camera: smile, look away, gesture. Up to 300 characters.
The price depends on the audio length: you pay for every second of the clip. The exact amount shows on the button once files are uploaded.