Want to sing over a song without the original singer, rehearse harmonies with your choir or hear an acapella more clearly? This free vocal remover works on a stereo audio file from your phone or computer and gives you an instrumental (vocals removed) or an acapella (vocals only). It runs in your browser, so nothing is uploaded.
How it works
- Choose a stereo song or beat from your phone or computer. Nothing is uploaded.
- Your browser splits the audio into thousands of short slices and compares the left and right channels in each frequency band.
- Sound that is identical on both sides (the centre of the mix, where lead vocals usually sit) is estimated and removed for the instrumental, or kept on its own for the acapella.
- Bass and kick drum below about 120 Hz are left alone by default, so the beat keeps its weight.
- Preview the result, then save it as MP3 or WAV.
Frequently asked questions
Will it remove the vocals completely?
Usually not completely. In our tests on a synthetic mix the centred voice dropped by about 15 dB, which sounds much quieter but still faintly audible. On real songs, reverb and echo on the voice are spread across both channels, so a ghost of them remains, and backing vocals that are panned to the sides stay in. For singing practice and karaoke this is often enough; for a clean commercial instrumental, ask the producer for the beat.
Why does my result still have vocals, or sound thin?
The remover can only find sound that sits in the centre of a stereo mix. If the song is mono, if the vocal is panned or heavily processed in stereo, or if the file has been converted many times, separation gets weaker. Anything else in the centre, often the snare, a lead synth or a solo instrument, is reduced too. Try the Gentle or Strong setting and keep "Keep bass and kick" switched on.
Is this AI?
No. The free tool uses signal processing (comparing the left and right channels in each frequency band), which is fast, private and works on phones. AI models give cleaner stems but are large downloads and slow on most phones, so we plan to offer them in Pro, processed in the cloud.
Can I use any song?
Only audio you own or have permission to use, such as your own recordings, beats you have licensed, or royalty-free and Creative Commons tracks whose licence allows edits. Cutting a song or removing its vocals does not give you any rights to it, so don't post, sell or perform the result without permission from the rights holders. We don't download audio from YouTube, TikTok, Spotify or any other link.
Is my audio uploaded anywhere?
No. Your file is opened and processed inside your browser on your own phone or computer. It is never sent to Centertainment or anyone else, so unreleased songs and private voice notes stay with you. The page only downloads the tool itself (and a small MP3 encoder file) from our site.
Is Audio Lab free?
Yes. Every tool on this page is free, with no sign-up, no watermark and no daily limit. Free files can be up to 10 minutes long (8 minutes for the vocal remover). A Pro version with high-quality cloud stem splitting, batch processing and longer files is planned; you can join the waitlist, and nobody is charged anything today.
Terms of use and copyright
- Only process audio you own or have permission to use: your own recordings, beats you have licensed, or tracks whose licence allows editing. Removing vocals, cutting or converting a song does not give you any rights to it.
- Don't publish, sell, stream or perform results made from other people's music without permission from the rights holders. Platforms may mute, block or claim such uploads, and you are responsible for how you use what you make.
- Audio Lab does not download or fetch audio from links (YouTube, TikTok, Instagram, Spotify, Audiomack, Boomplay or anywhere else) and does not host any music. You choose files that are already on your device.
- Your files are processed in your browser and never uploaded to Centertainment. We don't see, store or keep copies of your audio. Voice notes may contain other people's voices: respect their privacy.
- The tools are provided free and "as is". Results are automatic and can contain artefacts, so check them before you share, and keep your original files.
How the vocal remover works
In most studio mixes the lead vocal is placed dead centre: the same in the left and right speakers. Guitars, keys, pads and backing vocals are often spread to the sides. The oldest trick in karaoke is to subtract one channel from the other, which cancels the centre, but it also kills the bass and kick and turns the song mono.
Our version is smarter. It splits the song into thousands of short slices and frequency bands and, in each one, measures how alike the two channels are. Only the parts that match closely are treated as “centre” and removed, bass below 120 Hz is left alone, and the result stays in stereo. In our test on a synthetic mix with a known vocal, the centred voice dropped by about 15 dB while the panned instruments and bass came through almost untouched.
Honest limits
- It is not AI stem separation, so expect the voice to be much quieter rather than completely gone.
- Reverb and echo on the voice are spread across both sides, so a faint ghost often remains.
- Anything else in the centre (often snare, a lead synth or a solo instrument) is reduced too.
- Mono files, and songs where the vocal is panned or heavily widened, can’t be separated well.
Tips for better results
Use the best stereo source you have, not a WhatsApp forward or a screen recording. Start with Normal, try Strong if the vocal is still clear, and choose Gentle if the music starts to sound hollow. Keep “Keep bass and kick” switched on for most songs. Want the backing track for a song you wrote? Ask your producer for the stems: nothing beats the original files.
Please only process songs you own or have permission to use. Making an instrumental doesn’t give you rights to post or sell it. For more on making sing-along tracks, see our instrumental maker guide.