AI Hub

Vocal Remover

Separate vocals and instrumental tracks from any audio file — MDX-Net ONNX runs entirely in your browser.

Drop an audio file here

Supports MP3, WAV, M4A and more (stereo recommended)

Model: UVR_MDXNET_KARA_2 (~50 MB, downloaded once and cached).

All processing runs locally in your browser. Nothing is uploaded.

How to use

  1. Upload an audio file (MP3, WAV, M4A, etc.).
  2. Click Separate — the ~50 MB MDX-Net model downloads on first use.
  3. Listen to the Instrumental and Vocals previews, then download each track.

Examples

  • Extract the instrumental track from a song
  • Isolate vocals for karaoke or remixing
  • Separate backing music from a podcast or video clip

About the Model

Source: Blane187/all_public_uvr_models

License: MIT

MDX-Net ONNX (~50 MB) — browser wasm EP via onnxruntime-web; cached in Cache Storage. Input [1,4,2048,256] fixed shape; n_fft=4096, hop=1024, sr=44100.