Plugin Center · Audio Stem Separation
Audio Stem Separation
If you need to separate vocals, accompaniment, or multiple stems to prepare for dubbing, mixing, or review, start with this page.
On this page
- Best for vocal cleanup before dubbing and for post-mix preparation.
- Focus on the model notes, operating steps, and usage tips.
- Validate the separation quality on a short sample first.
Audio Stem Separation
Uses Demucs (htdemucs_6s) to separate vocals, drums, bass, guitar, piano, and other stems from audio or video.
Parameters and input
- Input modes: single file / multiple files (up to 50) / folder
- Supports audio and video formats such as MP3, WAV, MP4, MOV, MKV, and more
- Selectable stems to preserve: vocals, drums, bass, guitar, piano, other
- For video input, you can enable “Generate Video” to automatically export a video file with the replaced audio stem
Steps
- Open Plugin Center and choose “Audio Stem Separation”
- Choose the input mode and import audio or video material
- Select the target stems to keep, and optionally choose whether to generate a video output
- Start processing and review the separated results in the output folder
Usage note
The first run downloads the Demucs model (several hundred MB). Keep the network available and make sure FFmpeg is installed.