How does vocal removal work?

A trained AI model listens to the track and pulls the voice apart from the backing. It all happens on your own device — the audio never goes anywhere.

Does it remove vocals cleanly on every song?

It depends on how the track was mixed. Very wide stereo vocals may leave some residue. For a finer split by instrument, try 6stem.

A bit of the vocal is still there. What now?

Re-run on High Quality first — it takes longer, but the residue can change substantially. If it persists, try 6stem: the leftover vocal energy tends to get distributed into the guitar and other parts, where it is far less obvious. Do not reach for an EQ to notch the mids; the backing thins out before the voice does.

Do backing vocals and harmonies go too?

Usually, yes. The model was trained on human voice in general, not on lead versus harmony, so it does not distinguish between them — which also means harmonies you wanted to keep will leave with the lead. Everything vocal lands in a single track, so keeping one and dropping the other is not something this tool can do.

What do I get as a file?

WAV. You can save the backing on its own, the vocal on its own, or a balanced mix after adjusting the levels. If you want something smaller for a phone, the file converter on this site turns it into MP3 — also entirely in the browser.

Is it really free? Do I need to sign up?

It’s completely free with no account needed. Just open the page and pick a file.

Is the song I choose sent anywhere?

No. All separation happens entirely inside your browser.

Which file formats are supported?

MP3, WAV, M4A, AAC, FLAC and OGG, plus the audio track of video files.

Audio · processed on this device

Remove vocals from any song.

No upload — everything is processed inside your browser. Great for karaoke tracks and acapella extraction.

Free, no account
Stays on your device

About this process

Free in your browser · instant

One mix, separated into its individual parts

Vocal removal takes a finished stereo mix and lifts the singing out of it. Unlike the old approach of notching out the mid frequencies with an EQ, an AI model listens first and decides what is a voice and what is an instrument, so the backing keeps its body once the vocal is gone.

The old approach was centre cancellation: subtract one channel from the other. Vocals usually sit dead centre, so they disappear — along with everything else that sits centre, which is the kick, the snare and the bass. That is why the backing came out thin, and it never worked at all on a vocal spread across the stereo field. AI separation replaced that subtraction.

How clean it comes out depends on the mix. A vocal sitting centre with modest reverb separates well. Heavily doubled or wide-stereo vocals, and thick stacked harmonies, can leave traces of reverb or backing vocals in the instrumental. When that happens, 6stem often helps: the leftovers get distributed into separate parts you can mute individually.

Reaching for an EQ afterwards to notch out whatever residue is left is a poor trade — the backing loses more than the voice does. Re-running on a different model, or accepting that this particular track is a hard one, gets you further. How well a song separates is largely settled by how it was mixed, and there is no information to add after the fact.

Highlights

Separated by AI: A very different result from simply filtering frequencies — the backing keeps its body once the vocal is gone.
Never sent to a server: The audio you choose is processed in your browser and never uploaded.
Karaoke-ready backing: Vocals are removed while low and high frequencies keep their character, so the backing track is easy to sing to.
Adjust volume on the spot: Check the level of both vocals and backing track before you save.

Three simple steps

Open a file
Choose an MP3, WAV, M4A, FLAC or video with “Open file”, or drag it into the workbench.
Check settings and separate
Standard is enough for vocal removal. Choose High Quality for a cleaner split; pick Fast when you are in a hurry, or on a phone. Check the settings selected for this task, then press “Separate”.
Listen and export
Use Solo and Mute to listen to the parts you need, then choose “Export” to save WAV or MP3.

Vocal remover — frequently asked questions

How does vocal removal work?: A trained AI model listens to the track and pulls the voice apart from the backing. It all happens on your own device — the audio never goes anywhere.
Does it remove vocals cleanly on every song?: It depends on how the track was mixed. Very wide stereo vocals may leave some residue. For a finer split by instrument, try 6stem.
A bit of the vocal is still there. What now?: Re-run on High Quality first — it takes longer, but the residue can change substantially. If it persists, try 6stem: the leftover vocal energy tends to get distributed into the guitar and other parts, where it is far less obvious. Do not reach for an EQ to notch the mids; the backing thins out before the voice does.
Do backing vocals and harmonies go too?: Usually, yes. The model was trained on human voice in general, not on lead versus harmony, so it does not distinguish between them — which also means harmonies you wanted to keep will leave with the lead. Everything vocal lands in a single track, so keeping one and dropping the other is not something this tool can do.
What do I get as a file?: WAV. You can save the backing on its own, the vocal on its own, or a balanced mix after adjusting the levels. If you want something smaller for a phone, the file converter on this site turns it into MP3 — also entirely in the browser.
Is it really free? Do I need to sign up?: It’s completely free with no account needed. Just open the page and pick a file.
Is the song I choose sent anywhere?: No. All separation happens entirely inside your browser.
Which file formats are supported?: MP3, WAV, M4A, AAC, FLAC and OGG, plus the audio track of video files.

Practice with the extracted audio

Save the accompaniment, then change its key to suit your voice. Keep the original and listen to the changed audio before practicing.

Switch to “Pitch & tempo” in this workspace to use the audio with the current Solo and Mute settings. No file selection is needed. Export the separated parts as WAV or MP3 first if you want to keep them before further processing.

Change pitch and tempo

Guides for this task

Making your own backing track for a cover
No official off-vocal for the song you want to sing? Stem separation will make you one. The three-step process, plus two tricks for cleaning up whatever vocal is left behind.
How many semitones should you lower a song by?
The original sits too high and the chorus hurts. Lowering the key fixes that — until the low passages stop speaking. Where the useful range is, what it costs in quality, and why the order of operations matters.
How much faster is "tempo +1" on a karaoke machine?
Press tempo +1 and the song speeds up — by how much? Makers do not publish the step size, so here is how to measure it yourself and how to build a practice track at the same speed at home. Changing the tempo does not change the key.