No upload, 100% local, no account

Vocal remover (karaoke)

Drop a stereo track to cancel the centre channel and reduce the lead vocals. Everything runs in your browser; your file never leaves your device.

How Vocal remover (karaoke) works

Vocal remover produces a karaoke version of a song by applying centre-channel phase cancellation inside your browser. Drop a stereo track, pick stereo or mono output, and download the result. The work is done by ffmpeg compiled to WebAssembly, so your audio file is processed entirely on your device and never sent anywhere.

Be clear-eyed about the technique: this is not AI source separation. The tool subtracts the left channel from the right (and vice versa) using ffmpeg's pan filter, removing whatever is panned dead-centre, typically the lead vocal, but also centred kick, snare or bass. On some commercial mixes the result is surprisingly clean; on others, the vocal remains or obvious thinning appears. Results vary track by track.

Written by Bastien Sulyan

How to use Vocal remover (karaoke), step by step

  1. Drop a stereo MP3, FLAC or WAV file onto the upload area.
  2. Choose stereo output (wider, natural) or mono output (collapsed side signal, more consistent on some material).
  3. Click 'Remove vocals' and wait for the ffmpeg WebAssembly engine to finish.
  4. Download the resulting karaoke track directly to your device.

Common use cases

  • Backing-track practice: generate an instrumental version of a song to sing or play along with.
  • DJ prep: strip a vocal layer from a track you recorded yourself before remixing.
  • Karaoke night: make a quick sing-along version of a song your group already has as an MP3.
  • Music production reference: isolate the side content of a mix to study stereo width and instrument placement.

Frequently asked questions

Does the vocal remover use AI to separate instruments?

No. It uses centre-channel phase cancellation via ffmpeg's pan filter, not AI source separation. The filter subtracts the left and right channels from each other, which removes audio panned dead-centre. That is usually the lead vocal, but it also affects centred bass and drums. There is no machine-learning model involved.

Why does my song still have a faint vocal after processing?

Phase cancellation only removes what is mixed in the exact centre of the stereo image. If the vocal was double-tracked, panned slightly off-centre, or has reverb spread across the field, those elements survive. The effect is limited by how the original song was mixed, not a fault in the tool.

Why must the source file be stereo?

The technique works by computing the difference between the left and right channels. A mono file has a single channel, so there is no stereo difference to compute, and phase cancellation cannot do anything. Convert mono audio to stereo first if you need to try, though results are unlikely to be useful.

Does the ffmpeg WebAssembly engine receive my audio file?

No. ffmpeg is compiled to WebAssembly and runs entirely inside your browser tab. Your audio file is loaded into local memory, processed there by the pan filter, and the karaoke result is saved to your disk. Nothing is transmitted to Sunasty's servers or any third party at any point.

Is the output mono or stereo, and which is better?

Stereo output preserves the side-channel material in a standard two-channel file, which sounds wider and more natural. Mono output collapses the side signal to a single channel, which can sound more consistent on tracks where the stereo field is narrow. Try both and keep whichever sounds better for your specific track.

Can I process lossless formats like FLAC or WAV?

Yes. ffmpeg compiled to WebAssembly handles MP3, WAV, FLAC and most common audio containers. The processing time is proportional to the file duration; a typical 3-4 minute song finishes in a few seconds on a modern device.

Related tools

Keep everything local, explore complementary tools.

All audio tools