# SadTalker

> Talking-head generation from a single photo and an audio track — animates the face with natural head motion, eye blinks, and optional GFPGAN enhancement.

- **Provider**: replicate
- **Model ID**: cjwbw/sadtalker
- **Category**: lipsync_face_animation
- **Credits**: 91 per request
- **Pricing Type**: fixed

## API Endpoint

Base URL: https://api.core.today/v1

### Create Prediction
POST /predictions

### Get Status
GET /predictions/{job_id}

### Cancel
DELETE /predictions/{job_id}

## Authentication

Header: `X-API-Key: YOUR_API_KEY`

## Input Parameters

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

## Example Request

```json
{
  "model": "cjwbw/sadtalker",
  "input": {
    "pose_style": 0,
    "preprocess": "crop",
    "facerender": "facevid2vid",
    "source_image": "https://replicate.delivery/pbxt/IkgW9tngATq608Qf6haUXDpg81s5YBJfS9GaBiCFjdKXk4F5/art_1.png",
    "use_eyeblink": true,
    "size_of_image": 256,
    "expression_scale": 1,
    "use_enhancer": true,
    "still_mode": true,
    "driven_audio": "https://replicate.delivery/pbxt/IkgWA4bLoXpk5NwVsfOBzHh7MswfNLTgtf44Qr2gdOTOWvSX/japanese.wav"
  }
}
```

## Response Format

```json
{
  "job_id": "abc123",
  "status": "pending",
  "provider": "replicate",
  "model": "black-forest-labs/flux-schnell",
  "created_at": "2026-01-01T00:00:00Z",
  "result": null,
  "error": null
}
```

Status values: `pending`, `processing`, `completed`, `failed`, `cancelled`

## Usage Flow

1. POST /predictions with model and input → receive job_id
2. Poll GET /predictions/{job_id} until status is `completed` or `failed`
3. Result contains output URL(s) or data

## Output Type

url

## Tags

talking-head, lipsync, face-animation, avatar, sadtalker

## Documentation

https://replicate.com/cjwbw/sadtalker