How this vocal remover works
In most studio songs the lead vocal is mixed in the centre: it is equally loud in the left and right channels. Many instruments, such as guitars, keyboards and backing vocals, are placed more to one side, and stereo reverb is spread across both. The vocal remover uses this. It cancels what is identical in both channels, the centre, and keeps what is different, the sides. This method is called centre-channel cancellation.
A simple centre cancel also removes the bass and the kick drum, because they are usually in the centre too. That is why this tool protects the low frequencies: it removes the centre only above a chosen frequency and keeps the bass and kick below it. The result is a fuller backing track than with a basic phase-inversion trick.
The song is processed in your browser and is not uploaded.
Two modes
- Remove vocals: you get the song with the centre voice reduced, useful for karaoke, singing practice or a backing track for a cover.
- Extract vocals (centre): you get mainly what is in the centre, which is often the voice, plus some centre instruments. Useful for learning a melody or lyrics.
Honest expectations
Results depend strongly on how the song was mixed. This method works best on:
- stereo studio recordings with a clear lead vocal in the centre;
- pop, rock and many older recordings with wide stereo instruments.
It works poorly or not at all on:
- mono files, where both channels are the same, so everything is "centre";
- songs where the vocal has a lot of stereo reverb or doubled voices left and right; you will hear a quiet "ghost" of the voice;
- live recordings and some modern mixes where many instruments are also in the centre.
This is not an AI separation tool, so expect a reduction of the voice rather than a perfect silence. On the positive side, it is fast, works offline, and does not change the instruments that are not in the centre.
Tips for a better karaoke track
- Start from the best quality file you have. WAV or FLAC works better than a low-bitrate MP3, because MP3 encoding blurs the stereo image.
- Raise the bass protection frequency if the bass sounds thin, lower it if you still hear the voice in the low range.
- After removing vocals, a gentle cut around 1–4 kHz with the equalizer can hide what is left of the voice.
- To sing in a more comfortable key, change the pitch of the backing track afterwards.
You can also combine your own voice with the backing track in Mix Audio.
A quick way to test if a song is suitable: open it, switch to remove vocals and listen to the first chorus. If the voice drops clearly and the instruments stay full, the rest of the song will usually work too. If the voice is barely reduced, the song was probably mixed with the vocal spread across the stereo field, and another song or a different version, such as an official instrumental, will give a better result.