How to Get an Acapella From Any Song (Free and Paid Ways, 2026)
An acapella is the vocal of a song on its own, with no drums, bass or music underneath. Producers use them for remixes, bootlegs, mashups and DJ edits, and singers use them to study a performance. Official acapellas exist for some songs, but for most you will have to pull the vocal out of the finished track yourself. This guide covers where to look first, the ways to extract a vocal, and how to clean it up so it sits in your project.
Use an acapella for your own sessions and practice. If you release or sell something built on someone else's vocal, you need permission from the rights holders.
Step 1: look for an official acapella
An official acapella is the clean studio vocal, so it always beats an extracted one. Places worth checking:
- The artist's or label's releases. Some singles ship with an acapella or instrumental version, especially in dance music.
- Remix contests. Labels and platforms sometimes release full stems for a contest, vocal included, with rules about how you can use them.
- DJ record pools. Pools aimed at working DJs often carry acapellas and intro edits, licensed for DJ use.
If none of those has it, extract it.
Step 2: extract the vocal
Vocal extraction uses a trained model to estimate the voice inside the mixed song. The result is often very good. It is still an estimate, so expect some bleed: a faint hi-hat, a ghost of the snare, a little of the reverb from the original mix.
Your DAW
Logic Pro has a Stem Splitter, FL Studio has stem separation, and recent versions of Ableton Live can separate stems too. Check which version you need, since this changes between releases. The advantage is that nothing leaves your DAW.
Free tools
Ultimate Vocal Remover and Demucs are free and widely used. Quality can be excellent. The cost is setup: you install the software, choose a model and wait while your computer works through the song.
Online services
Services such as LALAL.ai split a file you upload and charge by the minute or by plan. They are quick and need no install. Check their current pricing, and remember that you are uploading audio to someone else's server.
Premix
Premix does the extraction as part of a bigger job. You type the song's name or drop in a file, and it reads the key and BPM, saves a 24-bit WAV and a 320 kbps MP3, splits four stems (vocals, drums, bass and other) and puts everything in your project on the grid. On an Apple Silicon Mac with macOS 14 or newer the split runs on your computer and is free after a one-time install of its own engine. On Windows and Intel Macs it splits through LALAL.ai instead. It is $39 one-time for Mac and Windows, with a 14-day money-back guarantee. If all you ever need is one vocal from one song, a free tool may be enough.
Step 3: clean up the vocal
An extracted vocal usually needs a few minutes of work before it sounds like a studio acapella.
- Listen soloed, then in context. Bleed you hear soloed often disappears once your new drums play. Fix what you can still hear in the mix, not everything you can hear alone.
- Gate the gaps. Between phrases the extraction often leaves faint music. A gate, or simply cutting the silent parts of the clip, removes it.
- High-pass it. Most leftover kick and bass sits below the voice. A high-pass filter around 80 to 120 Hz, set by ear, takes it out without thinning the vocal.
- Tame cymbal bleed. Hi-hats that survive the split sit in the same range as "s" sounds. A de-esser or a narrow EQ cut in the upper mids can soften both.
- Decide about the reverb. The original mix's reverb often comes out with the vocal. You can keep it, reduce it with a de-reverb tool, or duck it under your own reverb. Adding your own space on top of a dry-ish vocal usually sounds more modern than fighting the old tail.
- Check the phase. If you also use the instrumental stems, play everything together against the original song. A thinner sum means one part needs its polarity flipped.
Step 4: put it in your key and tempo
An acapella only works once it matches your project. Find the song's key and BPM first (our guide to finding the key and BPM of a song covers this), then warp the vocal to your tempo and decide whether to move the vocal or your track. Small moves keep a voice natural: one or two semitones is usually fine, more than that starts to sound processed. The Camelot wheel shows which keys sit next to the vocal's own key, so you can often avoid pitching it at all.
Which way should you go?
| You want | Try |
|---|---|
| The cleanest possible vocal | An official acapella, if one exists |
| A vocal without leaving your DAW | Your DAW's built-in splitter |
| The best free extraction and you do not mind setup | Ultimate Vocal Remover or Demucs |
| A one-off with no install | An online service |
| The song, its key and BPM, and all four stems on your grid | Premix |
Watch Premix pull a song into a session in 3 minutes: https://youtu.be/ad3z2sjoR-4