🎬 ControlFoley — Video → Foley Audio

Generate synchronized Foley sound effects from video using Xiaomi Research ControlFoley.

Upload a video or select the official sample below.

You can use:

  • V2A — leave the prompt empty
  • TV2A — describe the sound you want
1 8
1 8
5 30
Official ControlFoley Sample
Input Video Sound Prompt Negative Prompt Duration (seconds) CFG Strength Inference Steps Seed

Usage

Pure Video → Audio

Leave the prompt blank.

Text-guided Video → Audio

Describe the expected sound.

Example:

the skateboard wheels scraping and grinding on the ground.

For the first test, use:

  • Duration: 8 seconds
  • Steps: 25
  • CFG: 4.5
  • Seed: 42

If you want a quicker test, reduce inference steps to 10–15.