dojotool
MP3AI

Vocal Remover Online

The AI separates vocals from music. You get an instrumental and the vocals as a separate track.

Files are processed on your device and never uploaded to a server.

  1. Choose a song
  2. Remove vocals
  3. Download the instrumental
20 seconds, a synthetic song with a CC0 license

How to use

  1. Click "Choose song" or drop a file onto the page. MP3, WAV, M4A and other formats your browser opens will work.
  2. Click "Remove vocals". The first time takes longer: the browser downloads a 67 MB neural network.
  3. Listen to the result and click "Download instrumental". You can download the vocals without music from the "Vocals only" card.

What you get

You get two MP3 files at 192 kbps. On a computer you can also download them as uncompressed WAV.

  • An instrumental: the music without vocals, for karaoke or practice.
  • Vocals without music (an acapella): for a remix or to learn the vocal part.

How to get a cleaner instrumental

How clean the result is depends on the original song. Use a studio recording in good quality: MP3 at 256 kbps or higher, WAV or FLAC. Distortion from a low-quality MP3 will stay in the instrumental.

Live recordings and songs with heavy reverb separate worse: the vocal reverb may stay in the instrumental.

If you can still hear the vocals, find the studio version of the song and process it.

FAQ

Is this vocal remover free?
Yes. There's no sign-up, and you can process as many songs as you like.
Is my song uploaded to a server?
No. The neural network runs in your browser on your device, and the song isn't sent anywhere. The browser downloads only the neural network from our server, once.
Can I remove vocals on my phone?
Yes. On a phone you can process files up to 20 MB. Keep the browser open while it works.
Can I keep the backing vocals?
No. The neural network separates all voices at once, including backing vocals and choir.
Which formats are supported?
MP3, WAV, M4A, OGG and other formats your browser opens. The instrumental and vocals are saved as MP3 at 192 kbps, 44.1 kHz, stereo. On a computer you can also download uncompressed WAV.
Why does it take longer the first time?
The first time, your browser downloads a 67 MB neural network. After that it stays on your device, and the next songs are processed without downloading it again.
Why is it fast on one device and slow on another?
If your device has a graphics card the browser can use, the neural network runs on it, and that's fast. Otherwise the processor does the work, which takes several times longer.
Which neural network removes the vocals?
UVR-MDX-NET Voc_FT from the Ultimate Vocal Remover project (by Anjok07 and aufr33, MIT license).