How to Remove Vocals From a Song (Free): Instrumental or Acapella

You want to rap over the beat of a song you love. Or you need a karaoke version, an acapella for a remix, or you just want to hear what's going on under the vocal of a track you're studying. Either way, you need to get the vocals out of a finished song.

Good news: in 2026 you don't need expensive software for that. AI stem separation has gotten good enough that you can pull a usable instrumental and acapella out of a song for free, sometimes in your browser and sometimes in a DAW you already own. This guide shows you three ways to do it, how to get the cleanest result, and what you can and can't do with the files afterwards.

Quick answer

• Fastest free option: drop the song into a free online vocal remover like ours. You get two WAV files back: an instrumental and a vocals only track.
• Already use Logic Pro or FL Studio? Both have built-in stem separation. In Logic it's called Stem Splitter, in FL Studio it's "Extract stems from sample".
• Using Audacity? It has a built-in Vocal Reduction effect, and on Windows and Linux a free AI plugin that does a cleaner job.
• Expect some leftovers. No tool removes vocals perfectly. Reverb, delay and backing vocals often stay faintly audible.

How does a vocal remover actually work?

There are two very different approaches, and knowing the difference saves you a lot of frustration.

The old trick: phase cancellation

In most mixes the lead vocal sits dead center, so it's identical in the left and right channel. If you flip the phase of one channel and add them together, everything that's identical in both sides cancels out. The vocal disappears.

The problem: the kick, the snare, the bass and the 808 usually sit in the center too. So they get cancelled or thinned out as well, and what's left is a hollow mono version of the beat. Reverb on the vocal is often wide, so that stays behind as a ghost. It's a fun trick, but rarely usable.

The modern way: AI stem separation

AI models are trained on thousands of songs where the separate tracks are known. They learn what a voice sounds like compared to drums, bass and other instruments, and they rebuild each part as its own track. That's why you can get a full instrumental with the drums and bass intact, plus a separate acapella. Nearly every tool worth using today works this way.

Method 1: free online vocal remover (no install)

The quickest route is our free vocal remover. It runs in your browser and your file is never uploaded: the separation happens on your own device. No account and no watermark.

1. Open the vocal remover and pick your song. It accepts MP3, WAV, M4A, FLAC and OGG, up to 10 minutes long.
2. Choose a model. "Best quality" uses Demucs v4, an open source model from Meta. "Fast" uses Open-Unmix, which is quicker but leaves more of the vocal behind.
3. Let it run. The first time, your browser downloads the model and caches it. According to the tool page, Fast takes about 30 seconds for a 3-minute song on their test computer, and Best quality can take a lot longer on a computer without graphics acceleration. On a Mac with graphics acceleration, Best quality took just over a minute.
4. Download both files. You get an instrumental without vocals and a vocals only acapella, both as WAV.

Tip: if you have the time, use Best quality. The difference is very noticeable in the instrumental, especially in busy choruses.

Method 2: stem separation in your DAW

If you already produce in Logic Pro or FL Studio, you can split the song right inside your project. Handy, because the stems land on their own tracks and you can start working right away.

Logic Pro: Stem Splitter

1. Drag the song into your project.
2. Control-click the audio region and choose Processing > Stem Splitter.
3. Pick the stems you want. Logic can pull out vocals, drums, bass, guitar, piano and other parts. For a plain instrumental, select only Vocals, and the rest gets combined into a submix you can also extract.
4. Click Apply. The stems show up as subtracks under the original, which gets muted.

According to Apple's Logic Pro user guide, Stem Splitter only works on Macs with an M1 or later Apple silicon chip, so not on Intel Macs.

FL Studio: Extract stems from sample

1. Drop the song into the Playlist as an audio clip.
2. Click the small Clip Menu icon in the top left corner of the clip.
3. Choose Extract stems from sample and pick the stems you need: drums, bass, instruments and vocals.
4. Each stem lands on its own Playlist track under the original.

According to Image-Line's FL Studio manual, the first time you use it FL Studio downloads the model, and an extraction can take a few minutes during which FL Studio doesn't respond. So don't panic if it looks frozen. Image-Line's edition comparison lists stem separation in all editions of FL Studio.

Method 3: Audacity (free, Windows, Mac and Linux)

Audacity gives you a few options, according to the Audacity team:

• Music Separation (AI): after installing the free OpenVINO plugin, go to Effect > OpenVINO AI Effects > Music Separation and choose the 2 stem mode for vocals and instrumental. This is the cleanest option in Audacity, but it's only available on Windows and Linux.
• Vocal Reduction and Isolation: built in on every platform, including Mac. Go to Effect > Special > Vocal Reduction and Isolation and choose Remove Vocals. Your track has to be stereo. This is closer to the old phase trick, so expect a thinner result.
• Manual invert: split the stereo track, invert one side and mix them back together. This is the phase cancellation trick from above, with the same downsides.

7 tips for a cleaner instrumental or acapella

1. Start with the best file you have. A WAV, FLAC or 320 kbps MP3 separates much cleaner than a low quality rip. Every bit of compression gives the AI less to work with.
2. Pick the best model, not the fastest. Waiting a bit longer usually pays off.
3. Know which songs separate badly. Mono recordings, old songs and very dense mixes are harder. So is a lead synth or horn in the same range as the voice: part of it can end up in the vocal track.
4. Clean up the acapella. Put a low cut on it to remove leftover kick and 808. A gate or careful volume automation between lines helps hide bits of the beat.
5. Clean up the instrumental. Hearing a faint ghost of the vocal? A small EQ dip in the midrange can make it less obvious. Once you rap or sing on top, most leftovers disappear under your own voice anyway.
6. Use only the parts you need. Often the intro or a chorus separates better than a verse. For a remix or a practice session, you don't always need the whole song.
7. Check it on headphones. Artifacts are easier to hear there, so you know what you're working with before you build on it.

What to do with your instrumental next

• Practice your verse or hook. Writing and rapping over a song you know well is a great way to work on your flow and breath control.
• Find the key and tempo. Analyze the original song (with drums, not the acapella) in Beat Lab to get the BPM and key. Handy when you set up Auto-Tune for practice.
• Make it fit your voice. Hook too high? Read how to transpose a beat to fit your voice.
• Make a remix or mashup. Got the acapella? Our guides on how to make a remix and how to make a mashup take it from there.

Can you release a song on an extracted instrumental?

This is where a lot of people get into trouble. Removing the vocals doesn't make the beat yours. The instrumental is still part of someone else's recording and composition, owned by the songwriters, publisher and often a record label. Using it for practice, karaoke at home or to study a mix is one thing. Putting your own song on it on Spotify, YouTube or SoundCloud is another: for that you need permission from the rights holders. The same goes for using an extracted acapella in your own release.

Without permission, your upload can get a Content ID claim, be taken down or have its earnings go to the rights holders instead of you. This is general information, not legal advice. If you're unsure about a specific song, ask the rights holders.

Want to release something of your own? Then start with a beat you're actually allowed to use. Read what free for profit beats really allow, or browse the NameOnTheBeat beats: you pick a license that spells out what you can do with the beat.

FAQ

Is there a completely free way to remove vocals from a song?

Yes. Our vocal remover is free with no account, and Audacity is free software. If you already own Logic Pro or FL Studio, their stem separation costs nothing extra.

Why can I still hear the vocals a little?

No vocal remover is perfect. Reverb and delay tails, backing vocals and ad-libs are the hardest to separate, and heavily compressed files make it worse. Try the best quality model and the best source file you have.

Can I remove vocals on my phone?

Our vocal remover runs in the browser on phones too. According to the tool page, the Best quality model is only used when your phone's browser supports graphics acceleration. Otherwise it falls back to Fast.

Can I remove vocals from a YouTube video or a Spotify song?

These tools work on audio files you have on your device. Streams on Spotify or YouTube aren't files you can drop in, and downloading them usually goes against those platforms' terms.

Does removing vocals affect the beat's quality?

A little. AI separation rebuilds the instrumental, so you can hear small artifacts, especially in hi-hats and cymbals. For practice it's fine. For a release you want the original instrumental or a licensed beat.

Ready to try it? Open the free vocal remover, or check out all our free music tools for producers, rappers and DJs.