---

name: "LatentSync" description: "Generate lip-synced videos by combining audio with existing video. Trigger: Use when the user wants to add lip sync to a video, or requests 'latentsync', 'lip sync', 'sync lips to audio', or 'dub video with audio'." allowed-tools: ["Bash"]

LatentSync

Synchronize lip movements in a video to match provided audio. Supports loop modes for cases where the audio is longer than the video.

Quick Start

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "latentsync",
    "version": "0.0.1",
    "input": {
      "video_url": "https://example.com/talking-head.mp4",
      "audio_url": "https://example.com/new-dialogue.mp3",
      "guidance_scale": 1
    }
  }'

Parameters

Parameter	Type	Default	Description
audio_url	string	(empty)	The URL of the audio to generate the lip sync for
guidance_scale	number	1	Guidance scale for the model inference
loop_mode	string	(empty)	Video loop mode when audio is longer than video. Options: pingpong, loop
seed	integer	(empty)	Random seed for generation. If None, a random seed will be used
video_url	string	(empty)	The URL of the video to generate the lip sync for

Output

• Type: video

Examples

Basic Lip Sync

{
  "model": "latentsync",
  "version": "0.0.1",
  "input": {
    "video_url": "https://example.com/speaker.mp4",
    "audio_url": "https://example.com/new-speech.mp3",
    "guidance_scale": 1,
    "seed": 123
  }
}

Looped Lip Sync for Longer Audio

{
  "model": "latentsync",
  "version": "0.0.1",
  "input": {
    "video_url": "https://example.com/short-clip.mp4",
    "audio_url": "https://example.com/long-narration.mp3",
    "guidance_scale": 1.5,
    "loop_mode": "pingpong"
  }
}

Related Models

• EchoMimic V3 - Generate talking head videos from images
• Kling V1 Pro AI Avatar - AI avatar video generation
• Character 3 - Audio-driven character video

Documentation

• EachLabs API Docs