People remove vocals for karaoke, for practicing an instrument over a favorite track, for making a cover, or just to hear what the band is doing underneath the singer. A few years ago this meant tricky audio tricks with mediocre results. Today AI models do most of the work.
There are two very different ways to do it, and they give different results. Knowing which one you are using helps you pick the right option and judge the quality fairly.
Two ways to remove vocals
The classic method is center removal. In many stereo songs the lead vocal is mixed equally into the left and right channels, so it sits in the center. Subtracting one channel from the other cancels everything that is identical in both, which removes the lead vocal. It is instant, but it also weakens other centered sounds such as the kick drum and bass, and it leaves reverb and stereo backing vocals behind.
The modern method is AI source separation. A neural network has learned what voices look like in a spectrogram and splits the song into separate parts, called stems, such as vocals, drums, bass and other instruments. It handles reverb and backing vocals much better and does not depend on how the song was panned.
Which method should you choose
Choose the method by what you need the result for. For a quick check or a very old stereo recording, center removal can be enough. For anything you will actually sing or play along to, AI separation is usually worth it.
| Method | Speed | Quality | Best for |
|---|---|---|---|
| Center removal (no AI) | Instant | Basic, some voice and reverb remain | Quick tests, old stereo mixes |
| Fast AI | Seconds to a minute | Good | Karaoke tracks and acapellas |
| Best quality AI | Slower, needs a strong computer | Cleanest, four stems | Remixes and careful practice |
Remove vocals step by step
The SoundBench Vocal Remover at soundbench.net/vocal-remover/ offers all three methods, and it runs on your own device, so your music is never uploaded.
- Open soundbench.net/vocal-remover/ and add your song. MP3, WAV, FLAC, M4A and video files all work.
- Keep Fast AI selected and click Remove vocals. The first time, the AI model downloads once, about 40 MB.
- When it finishes, play the result and switch the vocal stem off to hear the instrumental. Compare it with the original at any time.
- Choose an output format and click Export. You can also download all four stems at once if you used Best quality AI.
Get a better result
Start with the highest quality source you have. A FLAC or WAV file gives the AI cleaner information than a low bitrate MP3, and artifacts are easier to hear when the source is already compressed.
If a little of the voice remains, try Best quality AI, which splits the song into vocals, drums, bass and other instruments. You can also open the instrumental in the SoundBench Equalizer at soundbench.net/audio-equalizer/ and lower the narrow band where the leftover voice sits. Loud, heavily processed mixes are the hardest case for any remover, so expect small traces.
A note on copyright
Separating a song does not change who owns it. Instrumentals and acapellas are fine for private practice, karaoke at home and personal projects. If you want to publish or share a result, get permission from the rights holder first.
Frequently asked questions
Can I remove vocals from a song for free?
Yes. The SoundBench Vocal Remover is free, with no credits, watermarks or sign-up, and no limit on how many songs you process.
Why is some of the voice still audible afterward?
Reverb, backing vocals and heavily processed mixes are hard to separate completely. Try Best quality AI, start from a lossless file, or use the equalizer to reduce what is left.
Are my songs uploaded to a server?
No. The AI model is downloaded to your browser and runs on your device, so your audio never leaves it.
Related tools
Vocal Remover
Remove vocals with AI and get a clean instrumental.
Stem Splitter
Split a song into vocals, drums, bass and other.
Acapella Extractor
Isolate the vocals from any song with AI.
Equalizer
10-band EQ with presets and live preview.
Remove Background Music
Keep the voice and remove music from audio or video.