Accepts MP3 or WAV
AI Vocal Separation
MDX-Net neural network isolates and removes lead vocals.
Runs On-Device
The model downloads once, then every song processes locally.
100% Private
Your audio never leaves your device.
No Upload Wait
No server round-trip — processing starts immediately.
How It Works
1
Upload a Song
Drop your MP3 or WAV — any track, any genre.
2
Remove Vocals
The first run downloads the separation model once (~65MB), then every song after that processes without re-downloading.
3
Download Instrumental
Preview the result, then download the instrumental MP3.