cjwbw/sadtalker

Stylized Audio-Driven Single Image Talking Face Animation

Capabilities

No capability data available

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`driven_audio` *	string (uri)	Upload the driven audio, accepts .wav and .mp4 file	`—`	—
`source_image` *	string (uri)	Upload the source image, it can be video.mp4 or picture.png	`—`	—
`expression_scale`	number	a larger value will make the expression motion stronger	`1`	—
`facerender`	string	Choose face render	`"facevid2vid"`	facevid2vid pirender
`pose_style`	integer	Pose style	`0`	min: 0, max: 45
`preprocess`	string	Choose how to preprocess the images	`"crop"`	crop resize full extcrop extfull
`size_of_image`	integer	Face model resolution	`256`	256 512
`still_mode`	boolean	Still Mode (fewer head motion, works with preprocess 'full')	`true`	—
`use_enhancer`	boolean	Use GFPGAN as Face enhancer	`false`	—
`use_eyeblink`	boolean	Use eye blink	`true`	—

driven_audio required string

Upload the driven audio, accepts .wav and .mp4 file

source_image required string

Upload the source image, it can be video.mp4 or picture.png

expression_scale number

a larger value will make the expression motion stronger

Default: 1

facerender string

Choose face render

Default: "facevid2vid"

facevid2vid pirender

pose_style integer

Pose style

Default: 0 min: 0, max: 45

preprocess string

Choose how to preprocess the images

Default: "crop"

crop resize full extcrop extfull

size_of_image integer

Face model resolution

Default: 256

256 512

still_mode boolean

Still Mode (fewer head motion, works with preprocess 'full')

Default: true

use_enhancer boolean

Use GFPGAN as Face enhancer

Default: false

use_eyeblink boolean

Use eye blink

Default: true

Version: a519cc0cfeba Updated: 6/8/2026 159.4K runs