← Back

How it works Karaoke FAQ

Free AI Vocal
Remover

Remove the vocals from any song with AI and get the instrumental and the acapella — 2 free tracks a day, no sign-up

MP3, WAV or FLAC · up to 50 MB and 10 minutes · 25 credits a track

How it works

  1. Upload the song. Any MP3, WAV or FLAC up to 50 MB and ten minutes.
  2. The AI separates it. A model trained on isolated stems splits the mix into a music track and a vocal track — not the old phase-inversion trick, which only ever cancelled whatever sat dead centre.
  3. Mix it with two faders. Vocal down gives you a karaoke instrumental. Music down gives you a clean acapella. Anything in between is a rough remix, and you hear it before you commit to it.
  4. Download what you need. The vocals, the instrumental or both, as MP3 or WAV. Moved the faders? Download the mix exactly as you hear it.

What people use it for

Frequently asked questions

Is this vocal remover really free?
Yes. Without an account you can separate 2 songs a day for free — no sign-up, nothing to install, and the count resets every day. The line under the upload button shows how many you have left. Members separate as many tracks as their credits cover (25 credits a track), and every result is saved to their library. The other free tools have no daily limit at all, because they run in your browser.
How do I make a karaoke version of a song?
Upload it, let the AI separate it, then pull the vocal fader all the way down and press Save. What you export is the music on its own — a karaoke instrumental at the original tempo and key. There is a full walkthrough here.
Can I get just the vocals, without the music?
Yes — pull the music fader down instead. That gives you the isolated vocal: an acapella you can put over a new beat, resample, or use as a reference while you mix.
How clean is the separation?
Far cleaner than phase inversion, which could only cancel what sat dead centre and took the bass and snare with it. A modern separation model was trained on real isolated stems, so on most commercial mixes the result is clean enough to remix or perform over. Very dense mixes and heavy reverb still leave small artefacts — that is true of every separator, paid ones included.
Does removing the vocal lose quality?
Nothing is resampled or re-encoded on the way through, so a WAV export comes back at the source rate and bit depth — the loss people mean by “quality” here is not compression, it is separation artefacts: faint smearing where the model had to decide whether something was voice or an instrument doubling it. Every separator has them, paid ones included, and they are worst on dense, heavily limited masters. If you can, feed it a WAV rather than a stream rip — the model cannot recover detail the encoder already threw away.
What file types and lengths work?
MP3, WAV, FLAC, M4A and OGG, up to 50 MB and ten minutes. If your file is bigger, bounce it to a 192 kbps MP3 first — it makes no meaningful difference to the separation. The converter does it in your browser.
Can I release a remix made from separated stems?
The tool gives you the audio; it does not give you the rights. Separating somebody else’s record still needs clearance before a commercial release. Your own tracks, practice, and official remixes are the safe ground. Here is what stems are and how they are used.
What happens to my upload?
It is sent securely for processing and never published. If you are signed in, the separated tracks go to your own private library, ready to re-download for as long as your account is open.

One separation, six ways in

This page is the general one: both faders come up and you set them yourself. The others are the same AI separation opened on a particular job, so you can start where you already know you are going.

Related tools