UtilibyWritten by the Utiliby editorial team · Last updated October 13, 2026 · Rules and formulas reviewed October 2026
What this free vocal remover does
A vocal remover tries to take the singer out of a song and leave the backing track, which is what you need for karaoke, singing practice, a cover version or a dance routine. The best known paid services use machine learning models that split a song into vocals, drums, bass and other parts. They are impressive, but they upload your file, limit the free minutes, and charge for downloads.
This free online tool uses a classic technique that has been used for decades: phase cancellation. It works completely in your browser, so there is no upload, no account and no limit on how many songs you try. It also lets you preview the result, switch back to the original, keep the bass and drums, and choose how strongly the center is removed. It is honest about its limits: it works very well on some songs and only partly on others, depending on how they were mixed.
How to remove vocals in four steps
- Load a stereo song. MP3, WAV, M4A and FLAC all work. A mono file cannot be processed this way.
- Pick a preset. Standard is a good start. Use Keep bass if the result sounds thin, and Max removal for the strongest effect.
- Listen and compare. Press Play and switch between Result and Original. Change the strength and the edges until the vocal is as quiet as you can get without damaging the music.
- Download. Choose MP3 or WAV and click Process and download.
How phase cancellation works
A stereo song has two channels, left and right. When an engineer places a lead vocal in the exact center, the same vocal signal is in both channels. If you subtract the right channel from the left, anything that is identical in both cancels to silence, which removes the vocal. Anything that differs between the channels, such as a guitar on the left and a piano on the right, remains.
This idea goes back to the early days of stereo, and the mid and side description of a stereo signal comes from the work of Alan Blumlein in the 1930s. The center is the sum of the channels, and the side is their difference. This tool works on the side signal inside the chosen frequency range and leaves the rest of the mix as it was, which keeps the bass and the top end.
What phase cancellation cannot do
The method removes everything that is the same in both channels, not just voices. Lead vocals, but also kick drum, snare, bass guitar and any other center-panned instrument, all drop out. That is why a plain subtraction often sounds thin and empty. The tool reduces this by leaving the lowest and highest frequencies alone.
It also cannot touch sound that is not centered. Reverb on a vocal is usually stereo, so the echo of the singer stays. Double-tracked vocals and backing singers panned left and right stay too. Modern productions often use stereo widening, which makes center removal less effective. If you need a clean separation of a modern track, an AI stem splitter will do better, at the cost of uploading the file or installing software.
Tips for getting the best result
- Start with the Standard preset at 100 percent and listen before changing anything.
- If the instrumental sounds thin, raise the low edge to 250 Hz or use the Keep bass preset.
- If a bright cymbal or a sibilant sound is cancelled, lower the high edge to 5000 Hz.
- Lower the strength to 70 or 80 percent when a full cancel damages the music. A faint ghost of the vocal is often better than a hollow track.
- Use the highest quality source you have. Lossy files work, but a WAV or FLAC gives cleaner cancellation.
- Try the original mix, not a live recording, a mono remaster or a track with a heavy stereo widener.
Removing versus keeping vocals
Remove mode subtracts the center in your chosen band. Keep mode does the opposite: it keeps only the center in that band and reduces everything else. The second mode is a quick way to hear what is in the middle of a mix, to practice harmonies against a lead vocal, or to build a rough acapella for study. Since other center sounds, such as snare and bass, remain, and side-panned instruments are only reduced by about 6 dB, it is not a clean vocal stem.
Choosing the low and high edges
Human voices sit mostly between about 100 Hz and 8 kHz, with the fundamental of a singing voice often between 80 Hz and 1 kHz and the clarity of the consonants above 2 kHz. Cancelling only within this band leaves kick and bass below it and cymbals above it.
| Preset | Band | Good for |
|---|---|---|
| Standard | 150 Hz to 8 kHz | Most pop and rock songs |
| Keep bass | 250 Hz to 5 kHz | Mixes where bass and kick vanish |
| Max removal | 60 Hz to 14 kHz | Strongest removal, thinner sound |
What people use a vocal remover for
- Karaoke at home: a backing track for any song you own.
- Singing practice: hear the instruments without the guide vocal.
- Covers and tributes: build a rehearsal track for your own performance, where the rights allow.
- Dance and fitness routines: music without a distracting lead voice.
- Musicians: study the arrangement and the band without the singer.
- Language learners: reduce the music under speech in a song to hear the words, using Keep mode.
Troubleshooting
- The tool says the file is mono: use a stereo version of the song.
- The vocal is still loud: the vocal is probably not in the center. Try Max removal, or accept a partial result.
- Everything sounds thin: use the Keep bass preset or lower the strength.
- There is an echo of the vocal: this is reverb, which is stereo and cannot be cancelled by this method.
- The output is too quiet or too loud: the tool lowers the level if it would clip. Adjust volume with the Audio Volume Booster tool.
- MP3 export fails: check your connection or choose WAV.
Phase cancellation compared with AI stem separation
AI stem separators, such as the open-source Demucs, are trained on thousands of songs to recognize what a voice looks like in a spectrogram, so they can pull out a vocal even when it is not centered. They are heavy: they need a powerful computer or a cloud service. Phase cancellation is lightweight, instant and private, and it is perfect when a song fits its assumptions. It is a good first attempt, and it costs nothing to try. If the result is not good enough, you will have learned that the song needs a stem separator.
Copyright and fair use
Removing the vocal from a song does not give you any rights to it. The recording and the composition remain protected. Practicing or singing along in private is generally fine, but publishing a karaoke or cover track, streaming it, or using it in a video can require licences from the rights holders. Use your own music, royalty-free tracks or music you have permission to modify.
Privacy and how it works
Your browser decodes the song to samples, and the page applies a few filters and subtractions in memory with JavaScript. Nothing is uploaded, stored or logged. The MP3 encoder library loads from a public CDN the first time you choose MP3, but your music is not part of that request. Closing the tab removes the audio from memory.
Frequently asked questions
How do I remove vocals from a song for free?
Load a stereo song above, leave the preset on Standard, press Play to hear the result, then click Process and download. The tool subtracts the center of the mix in the vocal range, which removes a lead vocal that was recorded in the center. It runs in your browser, so nothing is uploaded.
Does it use AI?
No. It uses classic stereo phase cancellation, which is fast and private but less precise than AI stem separation. Paid AI services can separate vocals, drums and bass much more cleanly, especially on modern tracks. This tool is best when the vocal is dead center and the instruments are spread across the stereo field.
Why are some vocals still audible?
Phase cancellation only removes sound that is identical in both channels. Reverb, doubled vocals, backing vocals panned to the sides and stereo effects stay behind. That is normal, and it is the main limit of this method.
Why does the bass or kick drum disappear?
Bass, kick and snare are usually in the center, just like vocals, so subtracting the center removes them too. The Standard preset leaves everything below 150 Hz alone to keep bass and kick. If it still sounds thin, use the Keep bass preset, which protects up to 250 Hz.
Does it work on mono files?
No. A mono file has the same signal in both channels, so subtracting one from the other gives silence. The tool detects this and tells you. You need a real stereo recording.
Which songs work best?
Stereo recordings from the 1960s to the 1990s often work well, because vocals were mixed in the center and instruments were panned. Modern pop with wide vocal effects and mono-compatible mixes is harder.
Can I extract the vocals instead (acapella)?
Yes. Change the mode to Keep vocals. This keeps only the center of the mix in the chosen range. The result will contain any other center sounds too, such as bass, snare or kick, and instruments panned to one side are only lowered by about 6 dB instead of removed, so it is a rough isolation rather than a clean acapella.
What does Strength do?
At 100 percent the center is removed completely. Lower values remove part of it, which is useful when full removal damages the instruments. In Keep mode, the strength controls how much of the non-center sound is taken out.
What do the low and high edges mean?
Cancellation is applied only between these two frequencies. Sound below the low edge and above the high edge is left untouched. Raising the low edge keeps more bass. Lowering the high edge keeps more cymbal brightness.
Is my song uploaded to a server?
No. The file is decoded and processed on your device. The page downloads an open-source MP3 encoder from a public CDN the first time you choose MP3, but your audio is never sent anywhere.
What does the level change in band figure mean?
It shows how much the overall loudness inside your chosen band changed after processing. A big drop means most of that band was center content. A small drop means a lot of the band is stereo content, such as instruments panned left and right, which is expected.
Is the output mono or stereo?
The part of the music between the two edges becomes mono, because it is what is left of the side signal. Sound outside the edges stays in its original stereo. For karaoke use, this is normally fine.
Can I use the instrumental on YouTube or sell it?
Only if you own the song or have a licence. Removing the vocal does not change the copyright in the recording or the composition. Use it for practice, private karaoke or with permission.
Why does my result sound hollow or phasey?
Cancelling the center removes the part of the sound that carries body and focus. Try a lower strength, the Keep bass preset, or a narrower band.
Does it work on phones?
Yes, in a current browser. Processing is fast, but very long songs need more memory.
Is there a length limit?
No fixed limit. A five minute stereo song is processed in a second or two. Very long files depend on your device memory.
What can I print or save from the page?
Print / save summary prints the settings and the waveform, and Copy summary copies the settings as text. The processed audio file is the main download.
Sources and further reading
- Audacity Manual, Vocal Reduction and Isolation, a description of the same phase cancellation technique.
- A. D. Blumlein, UK Patent 394,325 (1931), Improvements in and relating to sound-transmission, sound-recording and sound-reproducing systems, the origin of mid and side stereo.
- R. Bristow-Johnson, Audio EQ Cookbook, for the biquad filter formulas used for the band limits.
- Défossez, A., Hybrid Spectrogram and Waveform Source Separation (Demucs), for an explanation of AI stem separation.
- US Copyright Office, Fair Use Index, general information about using protected material.
Sources were consulted in October 2026. Standards and rules can change, so always check the official page for your situation.