Vocal Remover & Karaoke Maker

How center-channel separation works

Stereo mixing follows conventions as much as it follows technique, and one of the most consistent is placing the lead vocal in the centre of the image — sent to left and right at equal level. That means comparing the two channels shows the vocal as the part that matches almost exactly, while guitars, keys, room mics, reverb returns and stereo synths are deliberately pushed away from that centre to give the mix width.

This tool relies on that pattern. It splits the track into overlapping 4,096-sample frames, converts each into the frequency domain, and for every frequency bin compares left and right using a normalised cross-power measure. A value near one means both channels carry the same signal at the same level — centred content. A value near zero means they disagree — something panned or spread wide. That per-bin value forms a soft mask, which is used to build a centre signal, and subtracting that centre signal from the original leaves the instrumental.

Using a soft mask instead of the older left-minus-right subtraction trick matters a lot. Straight channel subtraction does remove the vocal, but it also flattens the mix to mono and cancels anything else that was centred, which is why old-school karaoke tracks always sound thin. Masking in the frequency domain keeps the stereo image intact and only touches the frequencies where centred energy is actually present.

Tuning the controls for your source

Separation strength is the main control. Lower settings remove only the most clearly centred content, which protects drums and bass but leaves more vocal bleed behind. Higher settings widen the mask and catch more of the voice, at the cost of thinning out the kick, snare and bass — the other things that typically live in the centre of a mix.

The frequency limits serve a related purpose. Below about 120 Hz there's almost never any vocal content, only kick and bass, so leaving that range untouched keeps the low end fully intact. Above roughly 12 kHz you're mostly in cymbal and air territory, and processing it usually adds a swirling artefact without removing any audible vocal.

• Start at the default strength and only raise it if you can still hear the lead clearly
• If the drums lose punch, lower the strength or raise the low cut
• Modern loud masters separate worse than dynamic mixes — the limiter glues everything together
• Live recordings and mono-era records will not separate usefully at all

Common uses, and a rights reminder

The most obvious use is karaoke — stripping the vocal out of a song to sing over. Nearly as common is the reverse: keeping the isolated vocal to build an acapella for a remix, mashup, or practice reference. Producers use it to check how much of a mix is centred, singers study phrasing without the backing track, and teachers isolate a part for a lesson.

Whatever the use, separating a commercial recording doesn't grant any right to redistribute the output — it's still a derivative of a copyrighted master. Personal practice and private study are fine; publishing an acapella or selling a karaoke track requires permission from the rights holders.

You might also need

• Audio Cutter & TrimmerSelect a region of a track, keep or delete it, and smooth the edges with fades.
• Tempo & Pitch ChangerRetime a track or shift its key on their own, without the classic pitch-follows-speed effect.
• Audio AnalyzerGet a full technical read-out of a track: loudness, key, tempo, stereo width and spectral balance.